Vitamin D Receptor (VDR) and Cholesterol Homeostasis: Interplay of VDR Enzyme Targets and Vitamin D Deficiency

Holly P. Quach

A thesis submitted in conformity with the requirements for the degree of Doctor of Philosophy Department of Pharmaceutical Sciences University of Toronto

Vitamin D Receptor (VDR) and Cholesterol Homeostasis:

Interplay of VDR Enzyme Targets and Vitamin D Deficiency

Holly P. Quach

Doctor of Philosophy

Department of Pharmaceutical Sciences University of Toronto

2016 Abstract

Vitamin D deficiency is speculated to play a role in hypercholesterolemia. However, there has

been little molecular evidence to link the two until recent evidence identified the vitamin D

receptor (VDR) and its natural ligand, 1α,25-dihydroxyvitamin D3 [1,25(OH)2D3], as key regulators of cholesterol metabolism. In the liver, 1,25(OH)2D3-liganded VDR directly inhibited

the small heterodimer partner (Shp) to increase expression of cholesterol 7α-hydroxylase

(Cyp7a1), the rate-limiting enzyme for cholesterol metabolism to bile acids, a mechanism

independent of the farnesoid X receptor. Vitamin D deficiency was established in mice after 8

weeks of the D-deficient diet, which resulted in decreased levels of plasma and liver 1,25(OH)2D3, downregulation of hepatic Vdr and Cyp7a1, and elevation of Shp. Consequently, higher plasma and liver cholesterol levels were observed. Intervention with 1,25(OH)2D3 or vitamin D3 reversed the altered expression of these cholesterol-regulating genes and lowered cholesterol levels back to baseline levels. The correlations between liver cholesterol vs. liver 1,25(OH)2D3 and Cyp7a1

expression in mice were also found in human liver tissue, suggesting that the VDR could be a

potential therapeutic target for cholesterol lowering. However, the therapeutic utility of

1,25(OH)2D3 is limited by hypercalcemia and by its feedback control mechanisms. When given

exogenously, 1,25(OH)2D3 triggered inhibition of its synthesis enzyme (Cyp27b1) but induction

of its degradation enzyme (Cyp24a1) in a dose-dependent manner, events that are predicted by

compartmental and physiologically-based pharmacokinetic/pharmacodynamic modeling. These

models provide a foundation for predicting 1,25(OH)2D3 disposition for inter-species scaling and

for exploration of alternative dosing schemes and routes of administration to describe the dynamics

of 1,25(OH)2D3 in its new therapeutic roles. Overall, these findings highlight the importance of

vitamin D status and the VDR as regulators of hypercholesterolemia.

iii

Acknowledgments

First, I would like to express my gratitude and respect for my mentor, Dr. K. Sandy Pang. I am

grateful for her guidance, support, and infinite contributions to this work. Thank you for always

demonstrating what determination, enthusiasm, and commitment can accomplish. Your work ethic

is unparalleled and truly inspirational.

I am thankful to my advisory committee members, Drs. Ping Lee, Micheline Piquette-Miller, and

Reinhold Vieth, for their efforts and words of wisdom. I would also like to thank our collaborators,

Drs. Donald Mager, Carolyn Cummins, Geny Groothuis, and Albert Li for sharing their expertise.

I am further grateful to Dr. Geny Groothuis for her hospitality and providing me with the

opportunity to work in her laboratory at the University of Groningen in The Netherlands.

I would like to thank all of the past and present members of the Pang lab. This work could not

have been accomplished without all of your efforts, especially Edwin, Matthew, Paola, Joy, and

Stacie. I am so appreciative of the support and friendships that you have provided me with both inside and outside of the lab.

I would also like to acknowledge the financial support that I have received throughout the years from the Natural Sciences and Engineering Research Council of Canada, Ontario Graduate Student

Scholarship, and the University of Toronto Fellowship.

Finally, I would like to thank my parents for their love, patience, and support throughout the years.

I am eternally grateful for the sacrifices that you have made for me. Andrew, thank you for your endless support, encouragement, and laughter. I could not have done it without you by my side every step (or running stride) of the way.

Table of Contents

Acknowledgments...... iv Table of Contents ...... v Abbreviations and Terms ...... x List of Tables ...... xiii List of Figures ...... xv List of Appendices ...... xix 1 Introduction ...... 1 1.1 The Vitamin D Receptor (VDR) ...... 1 1.1.1 Role of the VDR ...... 2 1.1.1.1 Physiological roles ...... 2 1.1.1.2 VDR tissue distribution and species differences ...... 4 1.1.1.3 VDR regulation of enzymes, transporters, and nuclear receptors ...... 4 1.1.1.4 Polymorphisms ...... 5 1.1.2 VDR ligands...... 6 1.1.2.1 VDR ligand binding pocket ...... 7 1.1.2.2 1α,25-Dihydroxyvitamin D3 or 1,25(OH)2D3 ...... 7 1.1.2.3 Vitamin D analogs and alternate ligands of the VDR ...... 8 1.1.3 Pharmacokinetics/pharmacodynamics of 1,25(OH)2D3 ...... 10 1.1.3.1 Pharmacokinetics of 1,25(OH)2D3 ...... 10 1.1.3.2 Modeling ...... 11 1.2 Cholesterol and Bile Acid Homeostasis ...... 13 1.2.1 Metabolic pathway and regulation ...... 15 1.2.2 Hypercholesterolemia ...... 16 1.2.3 Role of the VDR in cholesterol ...... 17 1.3 Vitamin D Deficiency ...... 18 1.3.1 Pathogenesis and clinical associations ...... 19 1.3.2 Vitamin D deficiency and hypercholesterolemia ...... 20 1.3.3 Experimental models of vitamin D deficiency and hypercholesterolemia ...... 23 1.4 Significance of the VDR in cholesterol and deficiency ...... 27 2 Statement of Purpose ...... 28 2.1 Purpose of investigation ...... 28 2.2 Hypotheses ...... 29 2.3 Thesis outline ...... 30 3 Temporal Gene Changes in Tissue 1α,25-Dihydroxyvitamin D3, Vitamin D Receptor Target Genes, and Calcium and PTH Levels After 1,25(OH)2D3 Treatment in Mice ...... 31 3.1 Abstract ...... 32 3.2 Introduction ...... 32 3.3 Materials and methods ...... 35 3.3.1 Materials ...... 35 3.3.2 Pharmacokinetic study of 1,25(OH)2D3 in mice ...... 35 3.3.3 Plasma calcium and phosphorus analyses and PTH assay ...... 36 3.3.4 Tissue 1,25(OH)2D3 extraction and 1,25(OH)2D3 EIA ...... 36 3.3.5 Preparation of subcellular protein fractions of kidney and intestinal tissues ...... 37 3.3.6 Western blotting ...... 38 v

3.3.7 Quantitative real-time PCR ...... 38 3.4 Results ...... 39 3.4.1 Similar plasma and tissue decay of 1,25(OH)2D3 after single and multiple dosing of 1,25(OH)2D3 to mice ...... 39 3.4.2 Intestinal distribution and effects of 1,25(OH)2D3 on intestinal and colon VDR, Cyp24a1, and Trpv6 mRNA expression ...... 43 3.4.3 Induction of renal VDR, Cyp24a1, and Trpv6 and downregulation of Cyp27b1 mRNA were time and concentration dependent ...... 45 3.4.4 Induction of renal Mdr1 mRNA and P-gp protein by 1,25(OH)2D3 ...... 47 3.4.5 Temporal changes in intestinal and renal VDR, Cyp24, and Trpv6 protein expression vs. plasma calcium and PTH levels in single and multiple doses of 1,25(OH)2D3...... 47 3.5 Discussion ...... 51 3.6 Acknowledgments ...... 55 3.7 Statement of significance of Chapter 3 ...... 55 4 Vitamin D Receptor Down-Regulates the Small Heterodimer Partner and Increases CYP7A1 in a Mechanism Independent of the Farnesoid X Receptor ...... 57 4.1 Abstract ...... 58 4.2 Introduction ...... 58 4.3 Materials and methods ...... 60 4.3.1 Materials ...... 60 4.3.2 1,25(OH)2D3 treatment of mice in vivo ...... 61 4.3.3 Plasma analyses of calcium, phosphorus, ALT, and bile acids ...... 61 4.3.4 Determination of plasma and liver 1,25(OH)2D3 ...... 61 4.3.5 Quantitative real-time PCR ...... 62 4.3.6 Western blotting ...... 63 4.3.7 Plasma and liver cholesterol ...... 63 4.3.8 Statistics ...... 63 4.4 Results ...... 63 4.4.1 Plasma calcium is increased upon 1,25(OH)2D3 treatment ...... 63 4.4.2 Parallel changes in liver 1,25(OH)2D3 concentrations and temporal Cyp7a1 mRNA and protein expression in wild-type mice ...... 64 4.4.3 In hypercholesterolemic Fxr(-/-) mice, 1,25(OH)2D3 reduced plasma and liver cholesterol ...... 66 4.4.4 mRNA expression in liver, ileum, and kidney ...... 67 4.5 Discussion ...... 68 4.6 Acknowledgments ...... 69 4.7 Statement of significance of Chapter 4 ...... 70 5 Vitamin D Deficiency Triggers Hypercholesterolemia That is Reversed Upon Treatment with 1α,25-Dihydroxyvitamin D3 and Vitamin D3 in Mice ...... 71 5.1 Abstract ...... 72 5.2 Introduction ...... 73 5.3 Materials and methods ...... 75 5.3.1 Materials ...... 75 5.3.2 Human liver tissues ...... 75

5.3.3 Animal studies ...... 75 5.3.3.1 Diets ...... 75 5.3.3.2 Animal housing and tissue collection ...... 76 5.3.3.3 Establishment of the vitamin D-deficient mouse model ...... 76 5.3.3.4 Intervention with short-term 1,25(OH)2D3 ...... 76 5.3.3.5 Intervention with long-term vitamin D3 ...... 77 5.3.3.6 Bile acid pool size studies ...... 77 5.3.4 Analysis of 25(OH)D3, 1,25(OH)2D3, calcium, and PTH ...... 79 5.3.5 Quantitative real-time PCR ...... 79 5.3.6 Western blotting ...... 79 5.3.7 Determination of plasma and liver total cholesterol levels ...... 79 5.3.8 Determination of bile acids by LC-MS/MS ...... 79 5.3.9 Statistics ...... 80 5.4 Results ...... 81 5.4.1 Vitamin D deficiency mouse model with reduced levels of 25(OH)D3 and 1,25(OH)2D3...... 81 5.4.2 Vitamin D deficiency increased cholesterol due to reduced Vdr and elevated Shp expression, thereby reducing Cyp7a1 expression ...... 83 5.4.3 Human liver cholesterol levels are inversely related to 1,25(OH)2D3 and CYP7A1 expression ...... 86 5.4.4 Elevated cholesterol caused by vitamin D deficiency and HF/HC diet ...... 86 5.4.5 Intervention with 1,25(OH)2D3 ...... 87 5.4.6 Intervention with vitamin D3 ...... 89 5.5 Discussion ...... 92 5.6 Acknowledgments...... 96 5.7 Supplementary material ...... 97 5.8 Statement of significance of Chapter 5 ...... 99 6 Vitamin D Analogs for Cholesterol Lowering ...... 102 6.1 Abstract ...... 103 6.2 Introduction ...... 104 6.3 Materials and methods ...... 107 6.3.1 Materials ...... 107 6.3.2 Luciferase reporter assay for in vitro binding to VDR ...... 107 6.3.3 In vitro potencies of 25(OH)D3, 1α(OH)D3, and 1,25(OH)2D3 in HEK293 and Caco-2 cells, in absence or presence of ketoconazole ...... 108 6.3.4 In vivo potencies of vitamin D analogs in mice fed a Western diet ...... 109 6.3.5 Plasma analysis ...... 109 6.3.6 Determination of cholesterol concentrations in plasma and liver ...... 109 6.3.7 Determination of 1,25(OH)2D3 concentrations in plasma and liver ...... 110 6.3.8 Relative mRNA expression using quantitative PCR ...... 110 6.3.9 Quantification of relative protein expression using Western blotting ...... 110 6.3.10 Statistics ...... 110 6.4 Results ...... 111 6.4.1 Vitamin D analogs are less transcriptionally active than 1,25(OH)2D3 in vitro .111 6.4.2 Ketoconazole inhibition of bioactivation of 1α(OH)D3 decreases its effects on VDR target gene expression in Caco-2 cells ...... 113

vii

6.4.3 Basal expression of bioactivation enzymes in HEK293 vs. Caco-2 cells ...... 115 6.4.4 Effects of vitamin D analogs on body weight and plasma ALT, bile acids, calcium, and phosphorus concentrations ...... 116 6.4.5 1,25(OH)2D3 concentrations increase in plasma and liver following treatment with 25(OH)D3 and 1α(OH)D3 ...... 118 6.4.6 1α(OH)D3 treatment lowers plasma cholesterol in hypercholesterolemic mice .118 6.4.7 Plasma and liver cholesterol levels are inversely correlated with 1,25(OH)2D3 120 6.4.8 Effects of vitamin D analogs on hepatic nuclear receptors, transporters, and enzymes...... 121 6.5 Discussion ...... 123 6.6 Acknowledgments...... 125 6.7 Supplementary material ...... 126 6.7.1 Effects of vitamin D analogs on ileal and renal nuclear receptors, transporters, and enzymes ...... 126 6.8 Statement of significance of Chapter 6 ...... 130 7 Pharmacokinetic/Pharmacodynamic Modeling to Describe 1α,25-Dihydroxyvitamin D3- Mediated Cholesterol Lowering ...... 132 7.1 Abstract ...... 133 7.2 Introduction ...... 133 7.3 Materials and methods ...... 136 7.3.1 Materials ...... 136 7.3.2 In vivo pharmacokinetic study for i.v. dosing of 1,25(OH)2D3 ...... 136 7.3.2.1 Plasma 1,25(OH)2D3 analysis ...... 137 7.3.2.2 Quantitative real-time PCR ...... 137 7.3.2.3 Analysis of i.v. data by PKPD modeling ...... 138 7.3.2.3.1 Non-compartmental analysis ...... 138 7.3.2.3.2 Estimation of PD parameters for inhibition and induction ...... 138 7.3.2.3.3 Fitting of i.v. data with compartmental modeling...... 139 7.3.3 The PBPK(TM)-PD and PBPK(SFM)-PD models ...... 141 7.3.3.1 PBPK-PD modeling ...... 141 7.3.3.2 The PBPK(SFM)-PD model was superior to the PBPK(TM)-PD model for describing the i.p. and i.v. data ...... 144 7.3.3.3 The extended PBPK(SFM)-PD model ...... 146 7.4 Results ...... 147 7.4.1 Dose-dependent PK of the i.v. 1,25(OH)2D3 data according to the two-compartment model ...... 147 7.4.2 Cyp27b1 and Cyp24a1 mRNA expression for the i.v. data ...... 150 7.4.3 PD response versus concentration curves for the i.v. data ...... 152 7.4.4 Fitting of the PKPD model to i.v. 1,25(OH)2D3 concentration ...... 154 7.4.5 Indirect response model with inhibition and induction functions to explain i.v. 1,25(OH)2D3 data with the compartmental PKPD model ...... 158 7.4.6 Comparison of the two-compartment, PKPD, and indirect response models .....158 7.4.7 Use of PBPK(TM)-PD and PBPK(SFM)-PD models to simulate i.v. data ...... 160 7.4.8 Extending the PBPK(SFM)-PD model to the liver ...... 161

viii

7.5 Discussion ...... 163 7.6 Appendix ...... 168 7.7 Acknowledgments...... 171 7.8 Supplementary material ...... 171 7.9 Statement of significance of Chapter 7 ...... 172 8 Discussion and Conclusions ...... 174 References ...... 179 Appendix I – Supplementary material for Chapter 4 ...... 202 Appendix II – Supplementary material for Chapter 5 ...... 206 Appendix III – Supplementary material for Chapter 6 ...... 219 Appendix IV – Supplementary material for Chapter 7 ...... 222 Appendix V – Vitamin D Receptor Down-Regulates the Small Heterodimer Partner and Increases CYP7A1 to Lower Cholesterol ...... 225 Appendix VI – Physiologically-Based Pharmacokinetic-Pharmacodynamic Modeling of 1α,25-Dihydroxyvitamin D3 in Mice ...... 244 Appendix VII – List of Publications ...... 264

Abbreviations and Terms

γ Hill coefficient τ time delay function

1α(OH)D2 1α-hydroxyvitamin D2 or doxercalciferol

1α(OH)D3 1α-hydroxyvitamin D3

1,25(OH)2D3 1α,25-dihydroxyvitamin D3 or calcitriol

25(OH)D3 25-hydroxyvitamin D3 7-DHC 7-dehydrocholesterol ABC ATP-binding cassette ACAT acyl CoA:cholesterol acyltransferase AIC Akaike information criterion Asbt/ASBT rodent/human apical sodium dependent bile acid transporter AUC area under the curve BARE bile acid response element Bsep/BSEP rodent/human bile salt export pump C concentration Ca calcium CA cholic acid Caco-2 human epithelial colorectal adenocarcinoma cell line CaSR calcium-sensing receptor CDCA chenodeoxycholic acid CL clearance CVD cardiovascular disease Cyp/CYP rodent/human cytochrome P450 enzyme Cyp24a1/CYP24A1 rodent/human 24-hydroxylase Cyp27a1/CYP27A1 rodent/human 25-hydroxylase (mitochondrial) Cyp27b1/CYP27B1 rodent/human 1α-hydroxylase Cyp2r1/CYP2R1 rodent/human 25-hydroxylase (microsomal) Cyp7a1/CYP7A1 rodent/human cholesterol 7α-hydroxylase DBP vitamin D binding protein DCA deoxycholic acid DHCR7 7-dehydrocholesterol reductase

Emax maximal stimulatory effect

EC50 concentration producing 50% of maximal stimulatory effect EIA enzyme immunoassay

fQ fraction of intestinal flow perfusing the enterocyte region FC fold-change x

Fgf15/FGF19 rodent/human fibroblast growth factor 15/19 Fgfr4/FGFR4 rodent/human fibroblast growth factor receptor 4 Fxr/FXR rodent/human farnesoid X receptor Gapdh/GAPDH rodent/human glyceraldehyde-3-phosphate dehydrogenase HDL-C high-density lipoprotein cholesterol HEK293 human embryonic kidney 293 cell line HepG2 human hepatocellular liver carcinoma cell line Hmgcr/HMGCR rodent/human 3-hydroxy-3-methylglutaryl-coA reductase HNF hepatocyte nuclear factor

Imax maximal inhibitory effect

IC50 concentration producing 50% of maximal inhibitory effect ICP-AES inductively coupled plasma atomic emission spectroscopy IDL intermediate-density lipoprotein Insig-2 insulin-induced gene-2 i.p. intraperitoneal i.v. intravenous k elimination rate constant

ka absorption rate constant

KP tissue to plasma partitioning ratio LCA lithocholic acid LCAa lithocholic acid acetate LCAaMe lithocholic acid acetate methyl ester LDL-C low-density lipoprotein cholesterol LRH-1 liver receptor homolog-1 LXRα liver x receptor alpha MCA muricholic acid Mdr1/MDR1/P-gp rodent/human multidrug resistance protein 1 or P-glycoprotein Mrp/MRP rodent/human multidrug resistance associated protein MSC model selection criterion NR nuclear receptor Ntcp/NTCP rodent/human sodium taurocholate co-transporting polypeptide Oatp/OATP rodent/human organic anion transporting polypeptide Ost/OST rodent/human organic solute transporter P phosphorus p.o. oral PBPK physiologically-based pharmacokinetic PD pharmacodynamic PK pharmacokinetic

PPAR peroxisome proliferator-activated receptor PTH parathyroid hormone PXR pregnane X receptor Q tissue flow rate

Rsyn net synthesis rate RFLP restriction fragment length polymorphism RXR retinoid X receptor SFM segregated flow model for intestine Shp/SHP rodent/human small heterodimer partner Srebp-2/SREBP-2 rodent/human sterol regulatory element-binding protein-2 Sult/SULT rodent/human sulfotransferase

t1/2 half-life

tpeak peak time t-CA taurocholic acid t-αMCA tauro-α-muricholic acid t-βMCA tauro-β-muricholic acid t-ωMCA tauro-ω-muricholic acid TBS-T Tris-buffered saline with 0.1% Tween-20 TC total cholesterol TM traditional model for intestine Trpv/TRPV rodent/human transient receptor potential cation channel subfamily V Ugt/UGT rodent/human UDP-glucuronosyltransferase UV-B ultraviolet light-B V volume Vdr/VDR rodent/human vitamin D receptor VDRE vitamin D response element WSSR weighted sum of square residuals

xii

List of Tables

Table 1-1 Half-life of 1,25(OH)2D3 at different doses and routes of administration

Table 1-2 Clinical cross-sectional studies relating vitamin D and cholesterol

Table 1-3 Clinical intervention studies relating vitamin D and cholesterol

Table 1-4 In vitro models relating vitamin D and cholesterol

Table 1-5 In vivo models relating vitamin D and cholesterol

Table 3-1 Mouse primer sets for quantitative real-time PCR

Table 3-2 Non-compartmental estimates for 1,25(OH)2D3, after repeated doses of 2.5 µg/kg q2d x 4 i.p. to mice

Table 4-1 Mouse primer sets for quantitative real-time PCR

Table 4-2 Body weight and plasma calcium, phosphorus, ALT, and bile acid concentrations

Table 5-1 Percent composition of total bile acid pool sizes in mice

Table 5-2 Plasma and liver parameters for mice fed the normal or high fat/high cholesterol vitamin D-sufficient or D-deficient diets for a total of 11-weeks, with intervention with vitamin D3 for 4-weeks

Table S5-1 Donor information for human liver samples (untreated) used for correlations

Table S5-2 Composition of normal and high fat/high cholesterol diets

Table S5-3 Bile acid standards and parameters for LC-MS/MS

Table 6-1 Effects of vitamin D3, 25(OH)D3, 1α(OH)D3, and 1α(OH)D2 treatments on body weight and plasma ALT, bile acids, calcium, and phosphorus concentrations, compared with previously published 1,25(OH)2D3 data

Table 6-2 Effects of vitamin D3, 25(OH)D3, 1α(OH)D3, and 1α(OH)D2 treatments on hepatic mRNA expression and Cyp7a1 protein expression, compared with previously published 1,25(OH)2D3 data

Table S6-1 Mouse and human primer sequences for quantitative real-time PCR

Table S6-2 Effects of vitamin D3, 25(OH)D3, 1α(OH)D3, and 1α(OH)D2 treatments on ileal mRNA expression, compared with previously published 1,25(OH)2D3 data

Table S6-3 Effects of vitamin D3, 25(OH)D3, 1α(OH)D3, and 1α(OH)D2 treatments on renal mRNA expression, compared with unpublished 1,25(OH)2D3 data

xiii

Table 7-1 Assigned physiologic parameter values for PBPK-PD models

Table 7-2 Fitted parameters obtained from PBPK-PD models with TM and SFM nested within the models

Table 7-3 Non-compartmental pharmacokinetic parameters following administration of first i.v. doses of 1,25(OH)2D3 to mice

Table 7-4 Fitted pharmacokinetic parameters for first i.v. doses of 1,25(OH)2D3 according to the simple two-compartment model

Table 7-5 Pharmacodynamic parameters estimated from Cyp27b1 or Cyp24a1 fold change versus plasma 1,25(OH)2D3 concentration

Table 7-6 Simultaneous fitting of first dose or pooled first and third dose i.v. data to different models, with k12 scaled

Table S7-1 Fitted parameters obtained by modifying the liver compartment of the PBPK(SFM)-PD model to include Cyp7a1 mRNA or Cyp7a1 protein

Table A1-1 Body weights and plasma ALT at end of treatment period Table A2-1 Body weights of mice at the end of the model validation study

Table A2-2 Body weights and plasma calcium and phosphorus measurements of mice at the end of the short-term 1,25(OH)2D3 intervention study without calcium supplementation

Table A2-3 Body weights and plasma calcium and phosphorus measurements of mice at the end of the short-term 1,25(OH)2D3 intervention study with calcium supplementation

Table A2-4 Body weights of mice at the end of the bile acid pool size study with short-term 1,25(OH)2D3 intervention without calcium supplementation

Table A2-5 Body weights of mice at the end of the bile acid pool size study with short-term 1,25(OH)2D3 intervention with calcium supplementation

Table A2-6 Body weights of mice at the end of the long-term vitamin D3 intervention study

Table A2-7 Body weights of mice at the end of the bile acid pool size study with long-term vitamin D3 intervention

Table A4-1 Body weights for mice at the end of the single i.v. dose study

Table A4-2 Body weights for mice at the end of the repeated i.v. dose study

xiv

List of Figures

Figure 1-1 General structure of nuclear receptors

Figure 1-2 Feedback regulation of plasma 1,25(OH)2D3 and calcium

Figure 1-3 Vitamin D analogs and LCA derivatives

Figure 1-4 General schematic of two-compartment model and traditional PBPK model

Figure 1-5 Simplified cholesterol and bile acid biosynthetic pathway highlighting key intermediates, mediators, and end-products

Figure 1-6 Regulation of CYP7A1 by bile acids in the liver

Figure 1-7 1,25(OH)2D3-liganded VDR increases cholesterol metabolism and lowers cholesterol

Figure 3-1 Basal levels of 1,25(OH)2D3 in plasma, kidney, ileum, and bone, and tissue/plasma concentration ratios in untreated C57BL/6 control mice

Figure 3-2 Plasma and tissue 1,25(OH)2D3 (ileum, kidney, and bone) concentration-time profiles from a single dose or multiple doses (Days 0, 2, 4, and 6) of 2.5 µg/kg i.p. 1,25(OH)2D3 q2d x 4 to mice

Figure 3-3 Intestinal distribution of mRNA and effect of 1,25(OH)2D3

Figure 3-4 Temporal changes in renal mRNA expression of VDR, Cyp24a1, Trpv6, and synthetic enzyme, Cyp27b1, after a single dose or multiple doses of 1,25(OH)2D3

Figure 3-5 Temporal changes in renal mRNA and protein expression of multidrug resistance protein-1 or P-glycoprotein from a single dose or multiple doses of 1,25(OH)2D3

Figure 3-6 Distribution and temporal changes of VDR protein in duodenum, jejunum, ileum, colon, and kidney after multiple doses of 1,25(OH)2D3

Figure 3-7 Temporal changes in ileal and renal relative protein expression of Cyp24 after multiple doses of 1,25(OH)2D3

Figure 3-8 Distribution and temporal changes of Trpv6 relative protein expression in duodenum, jejunum, ileum, colon, and kidney after multiple doses of 1,25(OH)2D3

Figure 3-9 Temporal changes in plasma PTH concentration from a single dose or multiple doses of 1,25(OH)2D3 to mice

Figure 4-1 Correlation between liver 1,25(OH)2D3 concentration and hepatic and Cyp7a1 mRNA and protein expression in normal diet-fed wildtype mice

Figure 4-2 1,25(OH)2D3 treatment increases hepatic Cyp7a1 and decreases hepatic Shp mRNA expression and plasma and liver cholesterol in Western diet-fed Fxr(-/-) mice

Figure 4-3 Changes in mRNA expression in the liver, ileum, and kidney of normal and Western diet-fed Fxr(-/-) mice treated with 1,25(OH)2D3

Figure 5-1 Experimental designs for establishing the vitamin D-deficiency model, short-term intervention with 1,25(OH)2D3, and long-term intervention with D3 in mice

Figure 5-2 Reduced 25(OH)D3 and 1,25(OH)2D3 and elevated cholesterol levels in mice fed vitamin D-sufficient or vitamin D-deficient diets for 0- to 8-weeks

Figure 5-3 Changes in cholesterol-regulating genes in the liver and ileum with mice fed vitamin D-sufficient or vitamin D-deficient diets for 0- to 8-weeks

Figure 5-4 Changes in liver gene expression are correlated with liver 1,25(OH)2D3 levels in vehicle-treated mice fed normal vitamin D-sufficient or vitamin D-deficient diets

Figure 5-5 Cholesterol is correlated with 1,25(OH)2D3 and Cyp7a1 expression in the mouse liver tissues of vehicle-treated mice fed normal vitamin D-sufficient or vitamin D- deficient diets

Figure 5-6 Cholesterol is also correlated with 1,25(OH)2D3 and CYP7A1 expression in untreated human liver tissue

Figure 5-7 Short-term (1-week) intervention with 1,25(OH)2D3 decreases cholesterol in mice fed normal or high fat/high cholesterol vitamin D-deficient diets for a total of 8- weeks

Figure 5-8 Long-term (4-weeks) intervention with D3 decreases cholesterol in mice fed normal or high fat/high cholesterol vitamin D-deficient diets for a total of 11-weeks

Figure S5-1 Basal 1,25(OH)2D3 concentrations and plasma PTH and renal Cyp27b1 and Cyp24a1 expression in mice fed normal vitamin D-sufficient or vitamin D-deficient diets

Figure 6-1 Vitamin D analogs are less potent than 1,25(OH)2D3 for VDR activity and are blunted by ketoconazole in the cell-based systems

Figure 6-2 Relative mRNA expression of VDR target genes in Caco-2 cells treated with 25(OH)D3, 1α(OH)D3, and 1,25(OH)2D3 ± ketoconazole

Figure 6-3 Basal expression of CYP2R1 and CYP27B1 bioactivation enzymes in HEK293 vs. Caco-2 cells

Figure 6-4 1,25(OH)2D3 and cholesterol levels in plasma and liver in Western diet-fed mice treated with vitamin D analogs

xvi

Figure 6-5 1,25(OH)2D3 and cholesterol concentrations are inversely related

Figure 7-1 Fitting of i.v. 1,25(OH)2D3 data using a simple two-compartment model, PKPD model, and an indirect response model

Figure 7-2 Schematic presentation of the PBPK-PD models for 1,25(OH)2D3 kinetics in mice

Figure 7-3 Extension of the PBPK(SFM)-PD model to incorporate 1,25(OH)2D3-mediated activation of liver Cyp7a1, leading to increased metabolism of cholesterol

Figure 7-4 Plasma 1,25(OH)2D3 concentrations after i.v. administration of 2, 60 and 120 pmol doses versus basal levels

Figure 7-5 Relative mRNA expression for synthesis and degradation enzymes following single or repeated i.v. administration of 1,25(OH)2D3

Figure 7-6 Inhibition of Cyp27b1 and induction of Cyp24a1 by 1,25(OH)2D3

Figure 7-7 Fitting of single and repeated dose plasma 1,25(OH)2D3 i.v. data using a PKPD model

Figure 7-8 The indirect response model for simultaneous fitting of all i.v. data on plasma 1,25(OH)2D3 and FC of VDR target genes, Cyp27b1 and Cyp24a1

Figure 7-9 Observed versus simulated concentration-time profiles of 1,25(OH)2D3 after multiple i.v. doses (given every 2 days for 6 days) in plasma, liver, ileum, and kidney using the PBPK-PD models with nested TM and SFM for describing the intestine compartment

Figure 7-10 The extended PBPK(SFM)-PD model could adequately describe temporal Cyp7a1 mRNA and cholesterol levels in the liver

Figure 7-11 The extended PBPK(SFM)-PD model could also adequately describe temporal Cyp7a1 protein and cholesterol levels in the liver

Figure S7-1 FC in hepatic Cyp7a1 mRNA and protein expression versus liver 1,25(OH)2D3 concentrations following all four doses of 120 pmol 1,25(OH)2D3 i.p.

Figure 8-1 Summary of changes in cholesterol-regulating genes, cholesterol, and bile acid concentrations under vitamin D-deficient conditions

Figure A1-1 Combined 1,25(OH)2D3 and atorvastatin treatment increases hepatic Cyp7a1 and decreases liver cholesterol in Western diet-fed mice

Figure A1-2 Changes in mRNA expression in the liver, ileum, and kidney of normal and Western diet-fed mice treated with 1,25(OH)2D3, atorvastatin, or combined 1,25(OH)2D3 and atorvastatin

xvii

Figure A1-3 Incubation of human precision-cut liver slices in medium containing 0.1% DMSO or 100 nM 1,25(OH)2D3 for ATP assay and RNA or protein isolation

Figure A1-4 Effect of incubation on viability of human precision-cut liver slices following 24 h incubation with 100 nM 1,25(OH)2D3

Figure A1-5 Relative mRNA expression in human liver slices incubated for 24 h with 0.1% DMSO or 100 nM 1,25(OH)2D3

Figure A2-1 Correlation of plasma 1,25(OH)2D3 vs. 25(OH)D3

Figure A2-2 Short-term intervention with 1,25(OH)2D3 also decreases cholesterol in vitamin D- deficient diets supplemented with calcium

Figure A2-3 Relative Srebp-2, Hmgcr, and Dhcr7 mRNA expression in vitamin D-deficiency

Figure A2-4 Treatment with 1,25(OH)2D3 and vitamin D3 do not alter expression of ileal Fgf15

Figure A2-5 Changes in hepatic mRNA expression of nuclear receptors, enzymes, and transporters in short-term intervention with 1,25(OH)2D3

Figure A2-6 Changes in ileal mRNA expression of nuclear receptors and transporters in short- term intervention with 1,25(OH)2D3

Figure A2-7 Changes in renal mRNA expression in short-term intervention with 1,25(OH)2D3

Figure A2-8 Changes in hepatic mRNA expression in long-term intervention with D3

Figure A2-9 Changes in ileal mRNA expression in long-term intervention with D3

Figure A2-10 Changes in renal mRNA expression in long-term intervention with D3

Figure A3-1 Changes in mRNA expression of nuclear receptors, transporters, and enzymes in Caco-2 cells grown for 21 days and treated on day 18 for 3 days with 0.1% DMSO, 100 nM 1,25(OH)2D3, 100 nM 1α(OH)D3, or 10 µM LCAa

Figure A3-2 Changes in hepatic mRNA and Cyp7a1 protein expression and plasma cholesterol levels and mRNA expression in ileum and kidney following treatment with LCAa

Figure A3-3 Changes in hepatic mRNA and Cyp7a1 protein expression following treatment with LCAaMe

Figure A3-4 Changes in hepatic mRNA expression following 1,25(OH)2D3 treatment q3d x 19

Figure A4-1 Relative mRNA expression of renal and ileal Vdr following single or repeated i.v. administration of 1,25(OH)2D3

Figure A4-2 Changes in plasma calcium and renal and ileal Trpv6 mRNA expression following single or repeated i.v. administration of 1,25(OH)2D3 xviii

List of Appendices

Appendix I – Supplementary material for Chapter 4

Appendix II – Supplementary material for Chapter 5

Appendix III – Supplementary material for Chapter 6

Appendix IV – Supplementary material for Chapter 7

Appendix V – Vitamin D receptor activation down-regulates the small heterodimer partner and increases CYP7A1 to lower cholesterol

Appendix VI – Physiologically-based pharmacokinetic-pharmacodynamic modeling of 1α,25- dihydroxyvitamin D3 in mice

Appendix VII – List of publications

xix

Chapter 1

Introduction

1.1 The Vitamin D Receptor (VDR)

Nuclear receptors play a critical role in the regulation of transporters and enzymes that are responsible for the fate of both endogenous and exogenous compounds. All nuclear receptors share a common structure (Fig. 1-1A): a ligand-independent activation domain for interaction with cofactors (AF-1) at the N-terminus, a central DNA binding domain that targets the nuclear receptor to highly specific DNA sequences or DNA response elements, a hinge region, a ligand binding domain that differs for every nuclear receptor and allows for specific ligand binding and receptor dimerization, and a C-terminus domain for co-regulator interactions (AF-2) (Mangelsdorf et al., 1995). Nuclear receptors are located in the cytoplasm when unliganded or in the nucleus when bound to their DNA response elements while repressed by a corepressor complex. In the presence of a ligand, they can form homodimers or heterodimers, acting as transcription factors that bind to specific response element sequences on target genes. The nuclear receptor superfamily 1 (NR1) is known for its role in the regulation of transporters and enzymes that affect the balance of endogenous molecules and disposition of xenobiotics. One of its members, the vitamin D receptor (VDR; NR1I1), is traditionally known for its role in the regulation of calcium and phosphate homeostasis. However, more recently, the VDR was identified as a regulator of transporters and enzymes involved in drug disposition (Schmiedlin-Ren et al., 1997), effects that were mediated by 1α,25- dihydroxyvitamin D3 [1,25(OH)2D3], the natural and active ligand of the VDR. To exert its regulatory effects, 1,25(OH)2D3-liganded VDR heterodimerizes with the retinoid X receptor (RXR; NR2B1) and acts on vitamin D response elements (VDRE) in the promoter region to modulate transcription of VDR target genes (Fig. 1-1B).

Figure 1-1. (A) General structure of nuclear receptors. (B) Activation of the vitamin D receptor (VDR) by 1,25(OH)2D3 binding, heterodimerization with retinoid X receptor (RXR), translocation to nucleus to bind vitamin D response elements (VDRE) in the promoter region to modulate transcription of VDR target genes.

1.1.1 Role of the VDR

1.1.1.1 Physiological roles

Vitamin D is widely known for its physiological role in regulating calcium and phosphate homeostasis. Here, intricate communication is required between a number of organs, including bone, kidney, intestine, and parathyroid gland in order to maintain adequate levels of plasma calcium, phosphate, and 1,25(OH)2D3. 1,25(OH)2D3-liganded VDR acts as a transcription factor for a number of target genes that contribute to an increase in plasma calcium levels. In bone, 1,25(OH)2D3 activates osteoblasts and stimulates the maturation of osteoclasts to promote bone metabolism (Jones et al., 1998). In kidney, 1,25(OH)2D3 induces epithelial calcium channels (transient receptor potential cation channel subfamily V; TRPV5/6) and a cytosolic calcium binding protein, calbindin

D28K, to increase calcium reabsorption (Dusso et al., 2005). In the small intestine, 1,25(OH)2D3 induces TRPV6 for calcium absorption from the lumen (den Dekker et al., 2003) and increases the transport of calcium across the cell by inducing calbindin D9K (Dusso et al., 2005). Because TRPV5/6 play such an important role in calcium reabsorption, the regulation of these channels by

1,25(OH)2D3 has been extensively investigated (den Dekker et al., 2003). The mRNA expression of

TRPV5/6 in mice was significantly increased following treatment with 1,25(OH)2D3, suggesting that transcription is regulated by 1,25(OH)2D3, a finding that was supported by the identification of 2

VDREs in the human and murine TRPV5 and TRPV6 promoters (Hoenderop et al., 2001; Meyer et al., 2006). In addition to controlling TRPV5 expression transcriptionally, 1,25(OH)2D3 may also exert translational regulation (Hoenderop et al., 2001). Concurrently, 1,25(OH)2D3 also plays a role in phosphate homeostasis by mediating the increased transport of phosphate through the apical NaPi cotransporter in the kidney and intestine (Hildmann et al., 1982; Taketani et al., 1998).

Elevated plasma calcium levels are detected by the calcium-sensing receptor (CaSR) in the parathyroid gland which inhibits the release of parathyroid hormone (PTH) into the bloodstream. Consequently, inhibition of PTH causes a reduction of the synthesis enzyme (1α-hydroxylase or CYP27B1) and induction of the degradation enzyme (CYP24A1), thus reducing plasma

1,25(OH)2D3 levels (Jones et al., 1998). When plasma calcium levels are high, production of calcitonin, a polypeptide produced in the thyroid gland, is also increased and inhibits calcium absorption from the intestine and osteoclast activity in bone (Jones et al., 1998). These feedback controls act to maintain plasma 1,25(OH)2D3 and calcium levels in homeostasis (Figure 1-2).

2+ Figure 1-2. Feedback regulation of plasma 1,25(OH)2D3 and calcium (Ca ). Modified from Matthew R. Durk.

1.1.1.2 VDR tissue distribution and species differences

The presence of the VDR in numerous tissues in the body suggests that 1,25(OH)2D3 plays a greater physiological role than what was initially believed (Andress, 2006). Although the VDR is mainly localized in the bone, kidney, intestine, and skin (Sandgren et al., 1991), low but detectable levels were also found in mouse and human livers (Gascon-Barre et al., 2003; Chow et al., 2014) and different regions of the mouse brain (Durk et al., 2014). Indeed, activated VDR was identified to regulate cell proliferation and differentiation, as well as synthesis and secretion of cytokines and other hormones (Valdivielso et al., 2009). More recently, our laboratory has identified various novel roles of the VDR in the liver for cholesterol lowering (Chow et al., 2014) and in the brain for decreasing levels of amyloid-β (Durk et al., 2014).

Human, mouse, and rat VDR protein alignments have been examined previously (Kamei et al., 1995) and it was shown that the DNA-binding domain is highly conserved across species (100%), while the mouse VDR ligand-binding domain is 89 and 96% identical to humans and rats, respectively. However, the mouse hinge region between these two domains was only 55% identical to humans and 78% identical to rats (Kamei et al., 1995).

1.1.1.3 VDR regulation of enzymes, transporters, and nuclear receptors

Interests in the VDR as a regulator of drug disposition initially arose when it was first reported that

1,25(OH)2D3-liganded VDR induced intestinal cytochrome P450 3A4 (CYP3A4) expression to increase midazolam metabolism (Schmiedlin-Ren et al., 1997). It was not until several years later that the presence of a VDRE was identified in the human CYP3A4 gene (Schmiedlin-Ren et al., 2001). Further studies revealed that the role of the VDR was not limited to CYP3A4 and included induction of CYP2B6 and CYP2C9 (Reschly and Krasowski, 2006), and a potential role in the downregulation of human cholesterol 7α-hydroxylase (CYP7A1), the rate-limiting enzyme in cholesterol metabolism, in primary human hepatocytes and HepG2 cells (Han and Chiang, 2009;

Han et al., 2010). It was also demonstrated that 1,25(OH)2D3 regulated genes for its own detoxification in the rat intestine, inducing the expression of its degradation enzyme Cyp24a1, as well as Cyp1a1, Cyp3a1, and Cyp3a11 (McCarthy et al., 2005; Kutuzova and DeLuca, 2007). Additionally, both Cyp24a1-knockout mice and keratinocytes derived from Vdr-knockout mice failed to metabolize 1,25(OH)2D3, thereby confirming the importance of VDR-mediated induction 4

of CYP24A1 for the metabolism of 1,25(OH)2D3 (Masuda et al., 2005). In rat and human intestinal slices, exposure of 1,25(OH)2D3 induced expression of Cyp3a1, Cyp3a2, and CYP3A4 (Khan et al.,

2009). Additionally, 1,25(OH)2D3 treatment also increased the expression of phase II enzymes, such as human sulfotransferase SULT2A1 (Echchgadda et al., 2004), mouse Sult2a2 (McCarthy et al., 2005), and rat UDP-glucuronosyltransferase Ugt1a (Kutuzova and DeLuca, 2007).

The classic targets of 1,25(OH)2D3-mediated disposition pertain to the regulation of calcium and phosphate homeostasis, where VDREs have been identified in epithelial calcium channels and sodium-dependent phosphate cotransporters (Taketani et al., 1998; Weber et al., 2001; den Dekker et al., 2003). Meanwhile, the role of VDR in xenobiotic transporters has gained increasing interest over the last few decades. While treatment with 1,25(OH)2D3 was previously reported to induce P- glycoprotein (P-gp) (Schmiedlin-Ren et al., 1997; Thummel et al., 2001; Pfrunder et al., 2003; Fan et al., 2009), it was not until recently that a VDRE was identified in the human multidrug resistance protein 1 (MDR1) gene (Saeki et al., 2008). More recently, 1,25(OH)2D3 was also shown to induce the expression of murine multidrug resistance-associated protein Mrp3 (McCarthy et al., 2005), human MRP2, MRP4, and P-gp (Fan et al., 2009; Maeng et al., 2012), rat apical sodium dependent bile acid transporter (Asbt) (Chen et al., 2006), rat and human intestinal folate transporter (Eloranta et al., 2009), and human organic anion-transporting polypeptide (OATP1A2) (Eloranta et al., 2012).

While it is clear that the VDR regulates many enzymes and transporters involved in drug disposition, it also appears to play an important role in cholesterol and bile acid homeostasis through cross-talk with other nuclear receptors. VDR was found to inhibit the farnesoid X receptor (FXR; NR1H4) (Honjo et al., 2006) and liver X receptor alpha (LXRα; NR1H3) (Jiang et al., 2006), key regulators of cholesterol and bile acid homeostasis (explained further in Section 1.2). In addition, a VDRE was found for fibroblast growth factor 15/19 (Fgf15/FGF19 in rodent/human), a negative regulator of Cyp7a1 (Schmidt et al., 2010). Meanwhile, other studies found that the human peroxisome proliferator-activated receptor (PPARα and PPARδ) genes were also primary targets of the VDR (Dunlop et al., 2005; Sertznig et al., 2009).

1.1.1.4 Polymorphisms

Since the vitamin D endocrine system is involved in a wide variety of biological processes, variations in this system have been linked to several common diseases, including diabetes, cancer, 5

and cardiovascular disease (Valdivielso et al., 2009). In addition to epidemiological studies that associate circulating vitamin D hormone levels with disease (explained further in Section 1.3), genetic studies provide an opportunity to link epidemiological data with molecular evidence (Uitterlinden et al., 2004a). Genetic polymorphisms, which occur frequently in the population, have been included to explain the variations in risk of common diseases. Although recent studies have identified that many polymorphisms exist in the VDR gene, the influence of these polymorphisms on function remains largely unknown. In the early 1990’s, restriction fragment length polymorphisms (RFLPs) were discovered at the 3’ end of the VDR gene, including the ApaI, EcoRV, BsmI, TaqI, Tru9I, and FokI polymorphisms (Uitterlinden et al., 2004a). The multiple polymorphisms that exist in the VDR gene are likely to affect the VDR protein levels or function depending on the cell type, developmental stage, and activation status, thereby influencing a number of biological endpoints (Uitterlinden et al., 2004b). More recently, epidemiological studies have associated VDR polymorphisms with increased risk of certain diseases, including diabetes (Panierakis et al., 2009), Alzheimer’s disease (Wang et al., 2012), and various types of cancer (Mahmoudi et al., 2010; Perna et al., 2013; Mun et al., 2015). Furthermore, a few reports have associated VDR polymorphisms with circulating 25-hydroxyvitamin D3 [25(OH)D3] levels (Engelman et al., 2008; Lee et al., 2014), although polymorphisms in other genes, including DBP, CYP24A1, CYP2R1, CYP27B1, and 7-dehydrocholesterol reductase (DHCR7) were more commonly associated with 25(OH)D3 levels (Jolliffe et al., 2015).

1.1.2 VDR ligands

The discovery of non-traditional actions of 1,25(OH)2D3 has sparked a new interest in vitamin D therapy for the treatment of a number of diseases, including cancer, psoriasis, and hyperparathyroidism, as well as its use for immunosuppression to prevent autoimmunity and transplant rejection (Peleg, 2005). However, hypercalcemia has limited the therapeutic utility of

1,25(OH)2D3 for most of these applications. Recently, there has been emphasis on the development of vitamin D analogs (compounds that are structurally similar to 1,25(OH)2D3) and alternate ligands of the VDR (compounds that are structurally dissimilar to 1,25(OH)2D3) that retain the desired therapeutic efficacy without evoking the toxic side effects.

1.1.2.1 VDR ligand binding pocket

Therapeutically useful compounds must bind well to the VDR in clinically relevant tissues. The key structural portion of the vitamin D compound for VDR binding is the A-ring containing the 1α- hydroxyl group. For example, VDR binding of 25(OH)D3, precursor to 1,25(OH)2D3, is poor but hydroxylation of this metabolite at the 1α position induces high affinity for the receptor (Bouillon et al., 1995). X-ray crystallography of the human VDR ligand binding domain showed that

1,25(OH)2D3, the natural ligand, was anchored in the binding pocket with the A-ring facing the interior and the 1α-hydroxyl group forming hydrogen bonds with arginine 274 and serine 237 (Rochel et al., 2000). The hydroxyl group at the 3β position forms hydrogen bonds at serine 278 and tyrosine 143, however, structure-function studies have shown that these interactions are less important for binding to the VDR. Meanwhile, the ability of 1,25(OH)2D3 to bind and transactivate the VDR depends on 25-hydroxylation of the side chain. The 25-hydroxyl group forms hydrogen bonds with histidine 397 and histidine 305, residues that are responsible for binding of the ligand and stabilization of the VDR in a transcriptionally active conformation, respectively (Rochel et al., 2000). While these studies were performed on the human VDR, the crystal structure and ligand- receptor interactions are speculated to be similar in rodents due to the high similarity in the amino acid sequences of the VDR protein (Kamei et al., 1995). Elucidation of the structure of the VDR ligand binding pocket has allowed for the conception and synthesis of novel ligands that exhibit improved affinity for the VDR. The discovery of selective agonists of the VDR that are able to induce biological responses without triggering hypercalcemia are especially desired.

1.1.2.2 1α,25-Dihydroxyvitamin D3 or 1,25(OH)2D3

1,25(OH)2D3 is the natural and active ligand of the VDR. It is the product of sequential bioactivation of vitamin D3, which can be obtained from dietary sources or formed from 7-dehydrocholesterol (7- DHC) in the skin upon exposure to UV-B light (Jones et al., 1998). Over 99% of vitamin D metabolites are bound to plasma protein, mostly to the vitamin D binding protein (DBP) and also to albumin and lipoproteins to a lesser extent. Vitamin D undergoes 25-hydroxylation by CYP27A1 and CYP2R1 that are most abundant in the liver (Cheng et al., 2003), though ubiquitously expressed, to form 25(OH)D3. Plasma 25(OH)D3, the main circulating metabolite, is highly bound to DBP. The high lipophilicity and binding of vitamin D and its metabolites in plasma contributes to low levels

of the metabolites and reduced metabolism in tissues and a long circulating half-life (Safadi et al.,

1999). The 25(OH)D3-DBP complex is filtered through the glomerulus in the kidney and taken up by the endocytic receptor, megalin, which is present on the brush border of the renal proximal tubule cells (Safadi et al., 1999). Once in the tubular cells, DBP is degraded by legumain (Yamane et al.,

2002) and the free 25(OH)D3 undergoes 1α-hydroxylation by CYP27B1 to form 1,25(OH)2D3. The plasma levels of 25(OH)D3 and 1,25(OH)2D3 are regulated by the 24-hydroxylase degradation enzyme, CYP24A1, to produce 24,25-dihydroxyvitamin D3 and 1α,24,25-trihydroxyvitamin D3, respectively, for elimination (DeLuca, 1988). 1,25(OH)2D3 levels are further regulated by the CaSR-

PTH pathway (Figure 1-2). Due to this intricate relationship between 1,25(OH)2D3 and calcium, treatment with 1,25(OH)2D3 is often associated with the unwanted side effects of hypercalcemia, including hypercalciuria, vomiting, anorexia, constipation, polyuria, dehydration, and fever (Joshi, 2009). The delicate balance between efficacy and toxicity plays a major role in the utility of

1,25(OH)2D3 as a therapeutic. Because treatment with 1,25(OH)2D3 elicits increased plasma calcium levels, its traditional use is for the treatment of secondary hyperparathyroidism in patients with chronic renal failure where levels of calcium are low (Chan and DeLuca, 1979). However, more recently, 1,25(OH)2D3 has demonstrated promising anti-cancer properties (Deeb et al., 2007), for which its hypercalcemic effects are unwanted.

1.1.2.3 Vitamin D analogs and alternate ligands of the VDR

A major breakthrough in vitamin D-based therapeutics has been the development of compounds that retain many of the clinically relevant activities of 1,25(OH)2D3 but have much lower calcemic activities in vivo (Figure 1-3) (Brown and Slatopolsky, 2008). Vitamin D analogs are structurally similar to 1,25(OH)2D3 and have been developed and approved for the treatment of various diseases. Calcipotriol, the first non-calcemic analog approved for clinical use, was proven effective in treating psoriasis (Kragballe, 1995). In chronic renal failure patients who are unable to synthesize active

1,25(OH)2D3, secondary hyperparathyroidism was treated with 22-oxa-1,25(OH)2D3, 19-nor-

1,25(OH)2D2, and 1α-hydroxyvitamin D2 [1α(OH)D2] (Brown and Slatopolsky, 2008). Similar to

1α-hydroxyvitamin D3 [1α(OH)D3], a prodrug of 1,25(OH)2D3, 1α(OH)D2 is a prodrug of

1,25(OH)2D2. Both require bioactivation in the liver prior to exerting its effects. Both 1α(OH)D3 and

1α(OH)D2 have been identified as equipotent to 1,25(OH)2D3 in vivo and in vitro, respectively, although hypercalcemic effects were still present (Fan et al., 2009; Chow et al., 2013a). For vitamin 8

D analog therapy, the greatest potential for impact is in the treatment of various types of cancer. For example, 22-oxa-1,25(OH)2D3, calcipotriol, and EB1089 were all shown to inhibit proliferation of breast cancer cells in vivo (Abe et al., 1991; Colston et al., 1992a; Colston et al., 1992b). These analogs have the potential to act as either primary or adjunctive therapies for treating various cancers.

Meanwhile, alternate VDR ligands that are structurally dissimilar to 1,25(OH)2D3 have also been identified. Even though lithocholic acid (LCA), a secondary bile acid, displayed much lower affinity for the VDR compared with 1,25(OH)2D3 (μM vs. pM), it was still able to induce VDR target genes such as Cyp24 and Cyp3a (Makishima et al., 2002). More recently, LCA derivatives, including LCA acetate and LCA acetate methyl ester, are being explored as possible therapeutic VDR ligands that selectively activate VDR target genes without eliciting hypercalcemia (Ishizawa et al., 2008).

Figure 1-3. Vitamin D analogs and LCA derivatives from Brown and Slatopolsky (2008) and Ishizawa et al. (2008), respectively.

1.1.3 Pharmacokinetics/pharmacodynamics of 1,25(OH)2D3

In order to fully understand the therapeutic effects of 1,25(OH)2D3, its pharmacokinetic (PK) and pharmacodynamic (PD) properties need to be evaluated. Because the VDR is found throughout the body and plays a role in a number of physiological processes, it is crucial that 1,25(OH)2D3 enter these tissues or cells in order to directly exert its effects. Hence, the efficacy of 1,25(OH)2D3 as a therapeutic relies on its ability to bind to the VDR in clinically relevant targets.

1.1.3.1 Pharmacokinetics of 1,25(OH)2D3

Ample data exists in the literature on the PK of 1,25(OH)2D3. However, much of the information is equivocal. Some of the discrepancies in the reported PK parameters of 1,25(OH)2D3 exist due to species differences, inadequate sampling, or lack of consideration of PD effects. The half-life (t1/2) of 1,25(OH)2D3 appears to be dose- and route-dependent (Table 1-1). In humans, a 4 µg dose of

1,25(OH)2D3 administered intravenously (i.v.) or orally (p.o.) resulted in t1/2 of 25.9 and 28.2 h, respectively (Brandi et al., 2002), whereas a similar dose of 4.2 µg led to a shorter t1/2 of 16.5 h

(Salusky et al., 1990). A dose-dependent effect wherein a shorter t1/2 with a higher dose was noted when rats were administered 2 or 10 µg 1,25(OH)2D3 i.v. (Kissmeyer and Binderup, 1991). These shorter t1/2 values were a result of higher clearances with higher doses. In contrast, mice treated with single doses of 0.125 or 0.5 µg 1,25(OH)2D3 intraperitoneally (i.p.) did not exhibit a clear dose- dependency, with t1/2 values of 7.6 and 7.8 h, respectively (Muindi et al., 2004), while Chow et al.

(2013b) reported a t1/2 of 6.8 h following a much lower i.p. dose of 0.05 µg 1,25(OH)2D3. Meanwhile, there was a route-dependent difference in the t1/2 of 1,25(OH)2D3 in rats pretreated with 0.4 µg

1,25(OH)2D3 i.p. or p.o., where a longer t1/2 was observed in rats that were administered

1,25(OH)2D3 p.o. (Vieth et al., 1990a). Vehicle-treated controls had a similar t1/2 for 1,25(OH)2D3 regardless of route of administration, suggesting that pretreatment of rats with 1,25(OH)2D3 altered the metabolism of 1,25(OH)2D3, a result that could be attributable to the existence of inducible enzymes that alter the synthesis (CYP27B1) and degradation (CYP24A1) of 1,25(OH)2D3 that are under control of the VDR (Vieth et al., 1990a). However, none of the studies reported in literature have examined these PD effects. Further, the unique PK profile of exogenously administered

1,25(OH)2D3 rests on the inhibition/induction of its own synthesis/metabolism, where levels will rise upon administration and then fall below baseline values over time. Bile acids could also follow

a similar pattern since increasing levels could inhibit its own synthesis/influx and promote its own degradation/efflux. A survey of the literature revealed that no other endogenous compound (e.g. cortisol, testosterone, thyroid hormone) appeared to follow this unique profile, albeit many studies failed to characterize the baseline values. Clearly, it is essential to consider these PD effects on the

PK of 1,25(OH)2D3 and vice versa when reporting PK parameters to avoid misinterpretation of results.

Table 1-1. Half-life of 1,25(OH)2D3 at different doses and routes of administration 1,25(OH) D Half-Life Species 2 3 Route Method Reference Dosea (h) i.v. competitive protein 25.9 4 µg Brandi et al., (2002) p.o. binding assay 28.2 human 4.2 µg i.v. competitive protein 11.4 Salusky et al., (1988) 4.2 µg i.v. binding assay 16.5 Salusky et al., (1990) 2 µg i.v. 3.8 HPLC Kissmeyer and 10 µg i.v. 2.3 Binderup (1991) rat i.p. 5.0 0.4 µg HPLC Vieth et al., (1990a) p.o. 10.4 0.125 µg i.p. 7.6 radioimmunoassay Muindi et al., (2004) mouse 0.5 µg i.p. 7.8 0.05 µg i.p. enzymeimmunoassay 6.8 Chow et al., (2013b) a based on a 70 kg human, 0.2 kg rat, and 0.02 kg mouse

1.1.3.2 Modeling

The development of a proper PK model that relates tissue concentration with effect will allow for

prediction of regulatory mechanisms that control the disposition of 1,25(OH)2D3. The compartmental and physiologically-based pharmacokinetic (PBPK) models can both be used to describe concentration and effect within tissues (Figure 1-4). In kinetic models, mass balance equations are written to describe the system mathematically and the factors affecting disposition are incorporated. For transport and metabolism kinetic processes, intrinsic clearances are used to describe influx and efflux across the cell membrane and metabolism. A typical PBPK model is often comprised of multiple compartments of discrete volumes that represent organs and tissues of interest that are linked by the blood circulation, providing insights into expected drug concentrations in major tissues. More complex scenarios like the segregated flow model (SFM) of the intestine that

considers a low proportion (5-30%) of blood flow to the active enterocyte region and the remaining flow to the serosal region can also be incorporated (Cong et al., 2000).

(A) (B)

F*Dosepo

Rsyn

k ka 12 Agut C1, V1 A2 k21

k10

Figure 1-4. General schematic of: (A) two-compartment model and (B) traditional PBPK model.

The interest in PD models has been steadily increasing due to the emphasis on translational and personalized medicine. The PD model describes the effect of a drug as a function of drug or metabolite concentration. PKPD models provide a link between these two principles that determine the relationship between dose and response. Principally, when a drug exerts an effect, a relationship will exist between plasma concentration and effect (Dingemanse and Appel-Dingemanse, 2007). However, pharmacological responses depend on numerous variables, including rate of drug input, distribution of the drug to its target, binding of the drug to its target, and activation or inhibition of homeostatic feedback mechanisms. It was first recognized that the intensity of many pharmacological effects is linearly related to the logarithm of dose (Levy, 1964). However, these simple linear models, wherein the drug acts centrally, represent the range in which the curvature

was neglected, and could not always capture the capacity or Emax parameter (Mager et al., 2003). In

this case, a sigmoidal Emax model or Hill equation better characterizes these relationships, where the pharmacological effects are seen immediately and directly related to the drug concentration and the

maximum effect is attributed to the drug (Dayneka et al., 1993). However, when the responses are not related to central elements and effects are not immediate but take more time to develop elsewhere, the observed response is not apparently related to plasma concentrations of the drug. When drug distribution to the site of action is needed in order to produce a biological effect, a biophase distribution model is used. Here, drugs exhibiting response delays are mathematically modeled to link the effect-compartment (or biophase) with plasma concentrations of the drug (Mager et al., 2003). Alternatively, an indirect response model is used to describe delays that may exist due to drug interactions that affect the fate of endogenous compounds and the subsequent effects that are mediated by these substances, where factors controlling the input (kin) or loss (kout) of the response variable may be either inhibited or stimulated (Dayneka et al., 1993). Additionally, the advent of mechanism-based PKPD models has given great impetus towards considering and incorporating the mechanism of action of the drug into establishing concentration-response relationships (Mager et al., 2003).

Hence, the development of PKPD or PBPK-PD models to describe the actions of exogenously administered 1,25(OH)2D3 would allow for the prediction of 1,25(OH)2D3-mediated biological actions, including potential therapeutic effects, at relevant target sites. In addition, knowledge gained from these studies would provide recommendations for the best dose and route of administration of

1,25(OH)2D3 for achieving therapeutic efficacy.

1.2 Cholesterol and Bile Acid Homeostasis

Cholesterol is an essential component of cell membranes that also acts as a precursor for a number of steroids, including bile acids and vitamin D. Cholesterol homeostasis is governed by the interplay between absorption, synthesis, and excretion or conversion of cholesterol to bile acids (Quintao et al., 1971). Thus, homeostasis is achieved through the coordinate regulation of several input and output events that are largely mediated by the liver. Pathways involved in cholesterol input include: uptake of serum cholesterol esters via low-density lipoprotein (LDL) receptor-mediated endocytosis; reverse cholesterol transport from peripheral tissues to the liver by uptake of high- density lipoprotein (HDL); absorption of dietary cholesterol from the intestine to the liver as chylomicrons by LDL receptor-mediated actions; and endogenous synthesis of cholesterol (Chiang, 13

2002), as depicted by Figure 1-5. The liver plays a key role in maintaining cholesterol homeostasis through feedback regulation mediated by the sterol regulatory element-binding protein 2 (SREBP- 2). SREBP-2 is the primary regulator of both the LDL receptor and 3-hydroxy-3-methylglutaryl coenzyme A reductase (HMGCR), the rate-limiting enzyme in cholesterol synthesis (Chiang, 2002). Meanwhile, pathways involved in cholesterol output include: assembly of cholesterol esters into very low-density lipoprotein (VLDL) and excretion into the circulation; conversion of VLDL to intermediary density lipoprotein (IDL) and LDL; and LDL receptor-mediated uptake of IDL and LDL into the liver and peripheral tissues (Chiang, 2002). Of the cholesterol catabolized, about 40% is excreted into bile for fecal elimination, 10% is used for synthesis of steroid hormones, and the remaining 50% is converted to bile acids (Chiang, 2002) through two major biosynthetic pathways. The classic (neutral) pathway produces roughly equal amounts of the primary bile acids, cholic acid (CA) and chenodeoxycholic acid (CDCA), and occurs in the liver where it is initiated by CYP7A1, the rate-limiting enzyme for cholesterol metabolism and synthesis of bile acids. The expression of sterol 12α-hydroxylase (CYP8B1) determines the ratio in which primary bile acids are formed, where higher expression favours the formation of CA (Li-Hawkins et al., 2002). Meanwhile, the alternative (acidic) pathway mainly produces CDCA and is initiated by sterol 27-hydroxylase (CYP27A1), an enzyme expressed in many tissues (Chiang, 2002). In humans, the primary bile acids formed are CA and CDCA, whereas in rodents, alternative hydroxylation gives rise to α-, β-, and ω- muricholic acids (α-MCA, β-MCA, and ω-MCA) in place of CDCA (Lefebvre et al., 2009). The rate of bile acid biosynthesis is controlled by enterohepatic circulation, where bile acids excreted in the liver are reabsorbed in the intestine and transported back to the liver (Hofmann, 1999). The bile acids are conjugated with taurine or glycine before excretion into bile where they are stored in the gallbladder and released into the intestine for digestion of fats after each meal. About 95% of bile acids are reabsorbed through ASBT-mediated uptake in the ileum, while the remaining 5% is lost in feces and replenished by de novo bile acid synthesis (Chiang, 2002).

acetate

3-hydroxy-3-methylglutaryl coenzyme A (HMG-CoA)

HMG-CoA reductase (HMGCR)

mevalonate

squalene

lanosterol

UV-B light desmosterol 7-dehydrocholesterol (7-DHC) vitamin D3

desmosterol reductase 7-DHC reductase (DHCR7) cholesterol classic (neutral) pathway alternative (acidic) pathway cholesterol 7α-hydroxylase (CYP7A1) sterol 27-hydroxylase (CYP27A1)

bile acids bile acids (CDCA, CA) (CDCA) Figure 1-5. Simplified cholesterol and bile acid biosynthetic pathway highlighting key intermediates, mediators, and end-products.

1.2.1 Metabolic pathway and regulation

Bile acids, the end-products of cholesterol metabolism, are toxic when they accumulate at high concentrations (Hofmann, 1999). Thus, the body tightly regulates bile acid synthesis through feedback control of cholesterol metabolism (Fig. 1-6). The promotor of the CYP7A1 gene contains a bile acid response element (BARE) that is highly conserved among species (Chiang, 2003). A number of nuclear receptors were previously identified to regulate cholesterol metabolism by altering CYP7A1 expression. When there is an excess of cholesterol in mice, oxysterols activate LXRα to increase Cyp7a1 expression for cholesterol metabolism, although LXRα does not appear to regulate CYP7A1 in humans (Chiang et al., 2001). The liver receptor homolog-1 (LRH-1) and hepatocyte nuclear factor-4α (HNF-4α) are transcription factors that also increase CYP7A1 expression (Chiang and Stroup, 1994; Crestani et al., 1998; Goodwin et al., 2000). Meanwhile, excess bile acids trigger a negative regulatory pathway to reduce CYP7A1 transcription by 15

activating the FXR, a bile acid sensor that controls bile acid and cholesterol homeostasis. FXR reduces bile acid synthesis via downregulation of the cholesterol-metabolizing enzyme, CYP7A1, through the induction of the small heterodimer partner (SHP; NR0B2), a repressor of CYP7A1 transcription that inhibits LRH-1 and HNF-4α (Goodwin et al., 2000; Lu et al., 2000). FXR also regulates bile acid levels in the liver by decreasing the sodium taurocholate co-transporting polypeptide (NTCP) and by induction of the bile salt efflux pump (BSEP) and MRP2, ATP-binding cassette efflux transporters on the canalicular membrane (Zollner et al., 2006) for control of both influx and efflux, respectively. In the ileum, FXR induces the fibroblast growth factor 15/19 (FGF15/19 rodent/human), a hormonal signaling molecule that represses CYP7A1 by interaction with fibroblast growth factor receptor 4 (FGFR4) in the liver (Inagaki et al., 2005). Bile acids can also bind and activate the pregnane X receptor (PXR; NR1I2) to inhibit Cyp7a1 expression (Staudinger et al., 2001).

Figure 1-6. Regulation of CYP7A1 by bile acids in the liver.

1.2.2 Hypercholesterolemia

Whole-body cholesterol levels are regulated by synthesis, absorption, excretion, and conversion of cholesterol into bile acids (Quintao et al., 1971). Several factors are thought to influence cholesterol 16

synthesis and absorption, including body weight, genetics, diet, and composition of the gut microbiome (Alphonse and Jones, 2016). Atherosclerosis, a known risk factor for cardiovascular disease (CVD), is a primary outcome of hypercholesterolemia where levels of LDL cholesterol (LDL-C) are elevated and levels of HDL cholesterol (HDL-C) are depressed (Kannel et al., 1979). Because excessive delivery of cholesterol to the periphery by LDL and VLDL leads to atherosclerotic plaque accumulation, LDL-C has been described as “bad cholesterol” and HDL-C as “good cholesterol” (Biggerstaff and Wooten, 2004). Traditional strategies for the prevention of CVD by reducing cholesterol levels in the bloodstream are focused on the manipulation of synthesis and absorption. For example, statins are a class of drugs that act by decreasing cholesterol synthesis via inhibition of HMGCR. Meanwhile, ezetimibe acts by inhibiting the Niemann-Pick C1 like 1 protein to reduce cholesterol absorption in the intestine (Altmann et al., 2004). A detailed review by Florentin et al. (2014) outlines a number of other treatment strategies, including both established and emerging drugs.

1.2.3 Role of the VDR in cholesterol

The VDR has surfaced to be another key regulator of cholesterol and bile acid homeostasis. A potential role of vitamin D in hypercholesterolemia was inferred when total cholesterol (TC) and

LDL-C levels in statin-treated patients were further reduced with vitamin D3 supplementation (Schwartz, 2009). Furthermore, VDR-knockout mice displayed elevated TC levels (Wang et al., 2009). In the rat, however, VDR expression in liver is virtually absent (Gascon-Barre et al., 2003), and the feedback control of the Fxr-Shp-Lrh-1 cascade on Asbt is non-existent (Chen et al., 2003; Chow et al., 2009). Hence, the rat is not the best model for studying VDR effects on cholesterol and bile acid homeostasis in the liver. Studies in human primary hepatocytes and HepG2 cells have suggested that the VDR inhibits CYP7A1 following treatment with 1,25(OH)2D3 (Han and Chiang, 2009; Han et al., 2010). However, some of these studies did not use time-matched controls that could confound interpretation of the data. Other in vitro studies showed that human VDR inhibits FXR (Honjo et al., 2006) and LXRα (Jiang et al., 2006), which would increase and decrease CYP7A1 expression, respectively. In mice, treatment with 1α(OH)D3, a vitamin D prodrug, resulted in upregulation of Cyp7a1 mRNA expression (Nishida et al., 2009). Meanwhile, treatment with a high dose of 1,25(OH)2D3 was shown to inhibit murine Cyp7a1 by increasing Fgf15 expression (Schmidt et al., 2010), whereas smaller doses of 1,25(OH)2D3 tended to activate the Vdr to suppress Shp and 17

increase Cyp7a1 expression and cholesterol metabolism (Chow et al., 2014). Thus, the VDR appears to be an important regulator of cholesterol metabolism and represents a potential therapeutic target for cholesterol lowering.

Figure 1-7. 1,25(OH)2D3-liganded VDR increases cholesterol metabolism and lowers cholesterol by downregulation of hepatic SHP (major mechanism) and possibly intestinal FGF15 (minor mechanism) in mice, as described by Chow et al. (2014).

1.3 Vitamin D Deficiency

The presence of the VDR throughout the body suggests that its activation, through maintaining vitamin D status to produce 1,25(OH)2D3, plays a critical role in health and disease. Vitamin D deficiency is alarmingly prevalent, and exists in approximately 30-50% of the general population in the U.S. (Lee et al., 2008). The current recommended dietary allowance is 600 IU/day (15 µg/day), although reports have indicated that doses up to 4000 IU/day (100 µg/day) are also considered safe

(IOM, 2011). Circulating 25(OH)D3 levels are used as an indicator to determine vitamin D status.

Although a consensus regarding the optimal level of 25(OH)D3 has not yet been established but was

suggested to range from 125-165 nM, most experts define deficiency as 25(OH)D3 levels less than 50 nM (or 20 ng/ml) (Lee et al., 2008; Sarkinen, 2011). Not surprisingly, vitamin D deficiency is associated with a number of skeletal diseases, including rickets, osteomalacia, and osteoporosis. However, more recently, a number of studies have associated vitamin D deficiency with non-skeletal diseases, including hyperparathyroidism, inflammation, hypertension, diabetes, cancer, and cardiovascular disease (Valdivielso et al., 2009). Hence, the role of the VDR and vitamin D status in the body appears to be of great importance.

1.3.1 Pathogenesis and Clinical Associations

Vitamin D deficiency is the consequence of inadequate sunlight exposure and/or insufficient dietary intake of vitamin D, and is further influenced by the latitude, season, age, skin pigmentation, obesity (Lee et al., 2008), and genetic variations of the synthetic and degradation enzymes that influence vitamin D status (Berry and Hypponen, 2011). The classical manifestation of vitamin D deficiency is in the form of rickets in children and osteomalacia in adults, conditions in which bone mineralization is impaired because of low plasma calcium and/or phosphate. Epidemiological and clinical evidence has implicated vitamin D deficiency as the cause of rickets and osteomalacia, while intervention studies have confirmed that vitamin D can effectively prevent or treat these conditions. Despite this knowledge, both rickets and osteomalacia remain a major problem around the world (Prentice, 2013) and especially prove to be a concern in the elderly population, where vitamin D deficiency is more prevalent and its consequences are more devastating (Gloth and Tobin, 1995).

More recently, there has been an abundance of studies describing the associations of vitamin D deficiency with a wide range of diseases, including various cancers and diseases of the cardiovascular, respiratory, neurological, musculoskeletal, and gastrointestinal systems (Reid and

Bolland, 2014). Many of these studies have inferred a causal relationship between low 25(OH)D3 levels and the associated disease, thus advocating vitamin D supplementation based on these observational results (Reid and Bolland, 2014). As vitamin D deficiency can be easily assessed and managed, a key point for ongoing studies is to assess the actual prevalence of the disorder rather than infer associations (Reid and Bolland, 2014); such studies will help clarify whether vitamin D deficiency is the cause of the disease or merely a consequence.

Animal models of vitamin D deficiency and the generation of VDR-knockout mice have allowed for detailed study of the mechanistic links between vitamin D deficiency and commonly associated clinical conditions. When compared with D-sufficient controls, mice fed the D-deficient diet exhibited enhanced tumor growths (Tangpricha et al., 2005; Ooi et al., 2010) and were also more prone to developing liver fibrosis (Zhu et al., 2015) and nonalcoholic steatohepatitis (Kong et al., 2014). In female mice fed D-deficient diets during pregnancy, fetal brain development was impaired (Hawes et al., 2015). In rats fed high fat diets, the introduction of D-deficient diets further exacerbated inflammation in nonalcoholic fatty liver disease (Roth et al., 2012). While these rodent models are useful for delineating the mechanistic links between deficiency and disease, we must be aware of the confounding factors that could limit the interpretation of these results. For example, adjustment of dietary calcium and phosphate levels is critical to avoid changes in mineral homeostasis that could confound conclusions drawn from deficiency studies (Stavenuiter et al., 2015). The use of VDR-knockout models are ideal because they describe the alterations that occur from absence of the VDR. VDR-knockout mice displayed osteomalacia (Bouillon et al., 2008), altered bone mass (Yamamoto et al., 2013), and increased serum cholesterol levels (Wang et al., 2009). Subtle abnormalities in immune and cardiovascular functions were also described, although their relevance to human disease remains unclear since cardiovascular function is normal in humans lacking VDR (Bouillon et al., 2008; Tiosano et al., 2013). Again, results must be interpreted with caution as mechanisms seemingly exist to adjust for the lack of VDR activity.

1.3.2 Vitamin D deficiency and hypercholesterolemia

Because insufficient sunlight exposure, a major cause of vitamin D deficiency, is found to be associated with elevated serum cholesterol levels (Grimes et al., 1996), it is plausible that vitamin D deficiency plays a role in hypercholesterolemia. The relationship between vitamin D status and cholesterol could be evaluated because both 25(OH)D3 levels and serum cholesterol (TC, LDL-C, and HDL-C) are routinely measured from blood samples taken in clinical studies. An examination of the literature revealed a common trend where low 25(OH)D3 levels were associated with increased TC and LDL-C levels, increased TC/HDL-C ratio and LDL-C/HDL-C ratio, or decreased HDL-C levels (Table 1-2). These trends held true in diverse populations that included people of different ethnicities and ages. Hence, there is growing evidence that supports the notion that vitamin

D deficiency plays a role in hypercholesterolemia. However, these cross-sectional studies could only provide associations and not causality.

Table 1-2. Clinical cross-sectional studies relating vitamin D and cholesterol

Subjects Key Results Reference 909 Finnish men, ↓ 25(OH)D3: ↑ TC, ↑ LDL-C Karhapaa et al., 45-70 years old ↑ 1,25(OH)2D3: ↑ HDL-C (2010) 39 Italian adults, Muscogiuri et ↓ 25(OH)D3: ↑ TC, ↑ LDL-C 41.4 ±12.4 years old al., (2010) 177 Spanish adults, Cutillas-Marco ↓ 25(OH)D3: ↑ TC, ↑ LDL-C 18-84 years old et al., (2013) 160 Brazilian adolescents, Oliveira et al., ↓ 25(OH)D3: ↑ TC, ↑ LDL-C 15-17 years old (2014) 88 Pakistani adults, ↓ 25(OH)D3: ↑ TC, ↑ LDL-C, Roomi et al., 20-40 years old ↑ TC/HDL-C (2014) 1484 Italian adults with coronary ↓ ↑ ↑ Verdoia et al., artery disease, 25(OH)D3: TC, LDL-C (2014) 20-80 years old 928 Kazak adults, Zhang et al., ↓ 25(OH)D3: ↑ LDL-C 30-75 years old (2014) 515 Saudi women in first trimester Al-Ajlan et al., ↓ 25(OH)D3: ↑ TC of pregnancy (2015) 307 Indian men, Patwardhan et ↑ 25(OH)D3: ↑ HDL-C 40-60 years old al., (2015) 659 Finnish adults, Pekkanen et al., ↓ 25(OH)D3: ↑ TC, ↑ LDL-C 66 ± 9 years old (2015) 782 Danish children, Petersen et al., ↑ 25(OH)D3: ↓ TC, ↓ LDL-C 8-11 years old (2015) 120 Italian obese children, Rusconi et al., ↑ 25(OH)D3: ↓ TC, ↓ LDL-C 10.2 ± 2.8 years old (2015) 136 Japanese men, Sun et al., ↓ 25(OH)D3: ↑ LDL-C/HDL-C 20-79 years old (2015) 326 Jordanian osteoporitic women, Yasein et al., ↓ 25(OH)D3: ↑ LDL-C 63.6 ± 8 years old (2015)

114 Singaporean adults, ↓ 25(OH)D3: ↑ LDL-C, ↑ LDL- Bi et al., (2016) 31.5 ± 12.4 years old C/HDL-C, ↑ TC/HDL-C

1172 Chinese adults, ↓ 25(OH)D3: ↑ TC, ↑ LDL-C, Li et al., (2016) 48 ± 11.9 years old ↓ HDL-C 209 obese American children, Lee et al., ↓ 25(OH)D3: ↑ TC, ↑ LDL-C 6-19 years old (2016) 21

In order to draw conclusions about causality, a number of intervention studies sought to increase

25(OH)D3 levels by exposure to UV-B light or through supplementation with vitamin D3. The findings from these studies produced mixed results (Table 1-3). A potential role of vitamin D in hypercholesterolemia was inferred when patients taking atorvastatin saw additional reductions in

TC and LDL-C levels following supplementation with vitamin D3 (Schwartz, 2009). Meanwhile, supplementation with calcium and vitamin D3 increased both 25(OH)D3 and HDL-C levels and decreased LDL-C (Schnatz et al., 2014). In contrast, other studies showed that supplementation with vitamin D3 had no impact or could even worsen the lipid profiles for some patients (Ponda et al., 2012; Salehpour et al., 2012). Clearly, the relationship between vitamin D status and cholesterol requires further exploration.

Table 1-3. Clinical intervention studies relating vitamin D and cholesterol

Subjects Treatment Key Results Reference 49 adults UV-B exposed group separated into 3 vs. control group in a UV-B exposed group: Carbone et al., groups based on tanning bed 2 times ↑ 25(OH)D3, ↓ LDL-C/HDL-C (2008) skin type per week for 12 weeks 45 patients 2000 IU vitamin D3 taking statins, Vitamin D3-treated group: p.o. daily and niacin, and ↑ 25(OH)D3, ↓ TC, ↓ LDL-C in Davis et al., increased as needed to omega-3 fatty both males and females, ↑ HDL-C (2009) achieve 25(OH)D3 of acid 50-60 ng/ml in males only supplementation 16 patients 800 IU vitamin D3 p.o. Vitamin D3-treated group: taking Schwartz (2009) daily for 6 weeks ↑ 25(OH)D3, ↓ TC, ↓ LDL-C atorvastatin 165 vitamin D- Vitamin D3-treated group: deficient 83 µg vitamin D3 p.o. ↑ ↑ Zitterman et al., 25(OH)D3, 1,25(OH)2D3, (2009) overweight daily for 12 months ↑ subjects LDL-C 45 vitamin D- 4000 IU vitamin D3 Vitamin D3-treated group: Longenecker et deficient HIV p.o. daily for 12 weeks ↓ TC, ↓ non-HDL-C al., (2012) patients 150 vitamin D- 50000 IU vitamin D3 Vitamin D3-treated group: deficient adults Ponda et al., p.o. per week for 8 ↑ LDL-C was correlated with with elevated (2012) weeks ↑ ↓ risk for CVD calcium and PTH

77 pre- Vitamin D3-treated group: 1000 IU vitamin D3 Salehpour et al., menopausal ↑ 25(OH)D3, ↑ TC, ↑ LDL-C, p.o. daily for 90 days (2012) obese women ↑ HDL-C 54 women with 50000 IU vitamin D3 gestational p.o. at day 0 and day Vitamin D3-treated group: Asemi et al., diabetes 21, monitored for 6 ↓ TC, ↓ LDL-C (2013) mellitus weeks 43 vitamin D- 0.5 µg 1,25(OH)2D3 1,25(OH)2D3-treated group: Bonakdaran et deficient type 2 p.o. daily for 8 weeks ↓ TC, ↓ LDL-C, ↓ HDL-C al., (2013) diabetics 50 vitamin D- deficient 50000 IU vitamin D3 Vitamin D3-treated group: women with p.o. once every 20 Rahimi-Ardabili ↑ 25(OH)D3, ↓ TC, ↓ VLDL-C et al., (2013) polycystic ovary days for 60 days syndrome 28 vitamin D- 16000 IU vitamin D3 Ramiro-Lozano Vitamin D3-treated group: deficient type 2 p.o. per week for 8 and Calvo- ↓ TC, ↓ LDL-C, ↓ non-HDL-C diabetics weeks Romero (2015) 600 post- 1000 mg calcium + Calcium/vitamin D3-treated ↑ ↑ Schnatz et al., menopausal 400 IU vitamin D3 p.o. group: 25(OH)D3 by 38%, (2014) women daily for 2 years HDL-C, ↓ LDL-C 1000 mg calcium per 118 vitamin D- Calcium/vitamin D3-treated day + 50000 IU deficient type 2 group: ↓ LDL-C, ↓ TC/HDL-C, Tabesh et al., vitamin D3 per week (2014) diabetics ↑ for 8 weeks HDL-C

1.3.3 Experimental models of vitamin D deficiency and hypercholesterolemia

In order to elucidate the underlying mechanisms that regulate cholesterol homeostasis under vitamin D-deficient conditions, in vitro and in vivo experiments are necessary. Seemingly, hypercholesterolemia could be a consequence of deficiency-related perturbations of either or both of the cholesterol synthesis and metabolism pathways. Several in vitro studies utilized vitamin D3 derivatives to determine its effects on the expression of cholesterol-regulating genes (Table 1-4).

Exposure of vitamin D3, 25(OH)D3, and 1,25(OH)2D3 to rat intestinal epithelial cells (IEC-6), human skin fibroblasts (GM-43), mouse peritoneal macrophages (J-774), and human hepatoma cells

(HepG2) showed varying results. While treatment with vitamin D3 and 25(OH)D3 inhibited

HMGCR activity in all cell lines, 1,25(OH)2D3 had no effect on HMGCR activity in IEC-6 cells, a biphasic effect in GM-43 and J-774 cells where only low concentrations were inhibitory, and a stimulatory effect in HepG2 cells (Gupta et al., 1989). In human promyelocytic leukemic cells (HL- 23

60) that are differentiated upon exposure to 1,25(OH)2D3, HMGCR activity is suppressed and activity of acyl CoA:cholesterol acyltransferase (ACAT), which catalyzes the transfer of fatty acid from acyl CoA to cholesterol, is increased to promote cholesteryl ester accumulation and foam cell formation in the macrophage (Jouni et al., 1995). In HepG2 cells, exposure to 1,25(OH)2D3 increased the expression of the insulin-induced gene-2 (Insig-2), an inhibitor of SREBP-2, thereby downregulating expression of both SREBP-2 and HMGCR to decrease intracellular total cholesterol

(Li et al., 2016). Furthermore, exposure to 1,25(OH)2D3 could also upregulate expression of the hepatic cholesterol efflux transporter, ABCA1, to promote synthesis of HDL-C (Yin et al., 2015).

In mouse and human hepatocytes, exposure to 1,25(OH)2D3 was shown to induce expression of Cyp7a1/CYP7A1, the cholesterol-metabolizing enzyme (Chow et al., 2014).

Table 1-4. In vitro models relating vitamin D and cholesterol

Model Conditions Key Results Reference Vitamin D3- or 25(OH)D3-treated : 0-5 µg/ml of vitamin ↓ HMGCR activity in concentration- Gupta et al., IEC-6 cells D3, 25(OH)D3, or dependent manner (1989) 1,25(OH)2D3 for 6 h 1,25(OH)2D3-treated: No effect on HMGCR activity Vitamin D3- or 25(OH)D3-treated: ↓ HMGCR activity in concentration- dependent manner 0-5 µg/ml of vitamin GM-43 and 1,25(OH)2D3-treated: Gupta et al., D3, 25(OH)D3, or J-774 cells Biphasic response where low (<1 (1989) 1,25(OH)2D3 for 6 h µg/ml) concentrations ↓ HMGCR but increasing concentrations returned activity to baseline Vitamin D3- or 25(OH)D3-treated: ↓ HMGCR activity in concentration- 0-5 µg/ml of vitamin dependent manner Gupta et al., HepG2 cells D3, 25(OH)D3, or 1,25(OH)2D3-treated: (1989) 1,25(OH)2D3 for 6 h ↑ HMGCR activity in concentration- dependent manner 5x10-8 M 1,25(OH)2D3-treated group: 1,25(OH)2D3 for 48 ↓ HMGCR activity, ↑ ACAT activity Jouni et al., HL-60 cells h to induce ↑ (1995) differentiation of to cholesteryl ester accumulation to cells promote foam cell formation

100 nM Mouse primary 1,25(OH)2D3-treated group: Chow et al., 1,25(OH)2D3 at 9 h hepatocytes ↓ Shp and ↑ Cyp7a1 mRNA (2014) (mRNA) 100 nM Human 1,25(OH)2D3 for 12 1,25(OH)2D3-treated group: Chow et al., hepatocytes h (mRNA) and 24 h ↑ CYP7A1 mRNA and protein (2014) (mRNA and protein) 1,25(OH)2D3-treated group: ↑ CYP27A1 to ↑ LXRα and ↑ 10 nM 1,25(OH)2D3 Yin et al., HepG2 cells ABCA1 mRNA and protein to for 24 h (2015) ↑ cholesterol efflux from liver to promote biogenesis of nascent HDL-C 10-8 and 10-9 M 1,25(OH)2D3-treated group: ↑ Insig-2 1,25(OH)2D3 for 6 h ↓ ↓ Li et al., HepG2 cells to SREBP-2 and HMGCR mRNA (2016) (mRNA) or 16 h ↓ (protein) and protein, intracellular TC

Recently, a growing number of in vivo studies have examined the effects of the VDR or vitamin D status on cholesterol (Table 1-5). In VDR-knockout mice, total cholesterol levels were elevated compared to those of wildtype mice (Wang et al., 2009), suggesting that the VDR may play a role in the determination of cholesterol levels. Meanwhile, Chow et al. (2014) showed that short-term

(1-week) treatment of mice with 1,25(OH)2D3 could decrease plasma and liver cholesterol levels of hypercholesterolemic mice through inhibition of Shp to increase Cyp7a1 expression. A common approach to study the effects of vitamin D deficiency is through the restriction of vitamin D in the chow. In apolipoprotein E (ApoE)- and LDL receptor (LDLR)-knockout mice fed vitamin D- deficient diets, increased serum cholesterol and plaque-associated lipids containing cholesterol were reported, respectively (Weng et al., 2013; Schmidt et al., 2014). When compared with wildtype mice that lack the cholesterol ester transport protein involved in transport of cholesterol esters from HDL to LDL, these knockout mouse models with elevated levels of LDL-C are more representative of the cholesterol profiles seen in humans (Hogarth et al., 2003). In rats, restriction of dietary vitamin D was shown to increase expression of Srebp-1c (Park et al., 2016), Srebp-2 and Hmgcr to increase TC, LDL-C, and LDL-C/HDL-C (Li et al., 2016). In a study by Yin et al. (2015), cholesterol levels were measured in Yucatan microswine fed a high cholesterol diet in combination with either a vitamin D-deficient, D-sufficient, or D-supplemented diet. When compared with the D-sufficient group, swine that were D-deficient displayed elevated LDL-C and decreased HDL-C, whereas the D-supplemented group showed lower LDL-C and increased HDL-C. Similar findings were reported 25

in a more recent study by the same group in swine that were subjected to coronary intervention (Gupta et al., 2016). Taken together, these results suggest that vitamin D deficiency could potentially cause hypercholesterolemia. However, none of these studies were able to demonstrate whether restoration of vitamin D levels in D-deficient animals, either by repletion in the diet or via treatment with vitamin D analogs, could rescue these effects. Moreover, these studies failed to identify a direct correlation between vitamin D status and cholesterol.

Table 1-5. In vivo models relating vitamin D and cholesterol

Animal Conditions Key Results Reference VDR-knockout 129S1 mice: ↑ TC, ↑ HDL-C in male mice VDR-knockout 129S1 or Wang et al., Mouse VDR-knockout NMRI mice: NMRI mice (2009) ↑ TC, ↑ HDL-C in male and female mice Macrophages from ApoE- ApoE-knockout mice fed a Weng et al., Mouse knockout mice fed D-deficient vitamin D-deficient diet (2013) diet: ↑ TC, ↑ free cholesterol C57BL/6 and FXR- knockout mice fed a high 1,25(OH)2D3-treated group: fat/high cholesterol diet for ↓ Shp mRNA to ↑ Cyp7a1 Chow et al., Mouse 3 weeks and treated with 2.5 mRNA, protein, and activity to (2014) µg/kg 1,25(OH)2D3 every ↓ plasma and liver cholesterol other day for 8 days Mice fed D-insufficient diet: LDLR-knockout mice fed a ↑ vascular calcification, Schmidt et al., Mouse vitamin D-insufficient diet ↑ (2014) for 32 weeks plaque-associated lipids consisting of cholesterol clefts

Rats fed D-deficient diet: Wistar rats fed vitamin D- Rat ↓ Vdr activity to ↓ Insig-2 to Li et al., deficient diet for 12 weeks (2016) ↑ Srebp-2 and ↑ Hmgcr

Goto-Kakizaki rats fed a high fat diet containing Low vitamin D3 group: ↓ ↑ Park et al., Rat vitamin D3 (low – 25 IU/kg, PPARα and Srebp-1c (2016) normal – 1000 IU/kg, high – mRNA 10000 IU/kg) for 8 weeks

Yucatan microswine fed When compared with D- high cholesterol and vitamin sufficient group: D-deficient (0 IU/day D3), D-deficient group had: Swine D-sufficient (1000 IU/day Yin et al., ↑ LDL-C, ↓ HDL-C (2015) D3), or D-supplemented D-supplemented group had: (3000 IU/day D3) diets for ↓ ↑ 48 weeks LDL-C, HDL-C Yucatan microswine fed high cholesterol and vitamin D-sufficient or D-deficient When compared with D- diets for 6 months and then sufficient + 1000 IU/day group, subject to coronary Gupta et al., Swine D-sufficient + 3000 IU/day intervention, followed by (2016) group had non-significant ↓ TC, supplementation with ↓ LDL-C, ↑ HDL-C vitamin D3 (1000 IU/day or 3000 IU/day) for another 6 months

1.4 Significance of the VDR in Cholesterol and Deficiency

The link between the VDR and cholesterol homeostasis is evident but the details remain relatively unknown. In clinical studies, vitamin D deficiency is often associated with hypercholesterolemia but these studies do not provide evidence for causality. While clinical studies have postulated an association, very few preclinical studies have sought a direct correlation between vitamin D status and cholesterol. In fact, there has been little molecular evidence to link the two until recently when the VDR was identified as a key regulator of cholesterol metabolism. In mice, 1,25(OH)2D3- liganded VDR was found to downregulate expression of Shp to increase Cyp7a1 and lower cholesterol levels (Chow et al., 2014). Meanwhile, in a rat model, vitamin D deficiency was shown to decrease Vdr activity leading to increased cholesterol synthesis via increased expression of Hmgcr (Li et al., 2016). While these data suggest that the VDR and vitamin D status may play a role in cholesterol homeostasis, the direct relationship between vitamin D deficiency and hypercholesterolemia remains undefined. Hence, understanding the mechanisms behind this potential relationship would provide great insights to explain the clinical associations that are often reported between low 25(OH)D3 and hypercholesterolemia.

Chapter 2

Statement of Purpose

2.1 Purpose of investigation

The VDR is traditionally known for its role in maintaining calcium and phosphate homeostasis in the body (Jones et al., 1998). Meanwhile, the identification of the VDR as a regulator of transporters and enzymes involved in drug disposition (Schmiedlin-Ren et al., 1997; Echchgadda et al., 2004; Fan et al., 2009; Kim et al., 2014) sparked interest in the VDR as more than merely a regulator of mineral homeostasis. A potential role of the VDR in cholesterol and bile acid homeostasis was first suggested when it was identified as a bile acid sensor involved in the detoxification of bile acids (Makishima et al., 2002), and then found to inhibit FXR and LXRα (Honjo et al., 2006; Jiang et al., 2006), nuclear receptors that are involved in cholesterol metabolism by regulation of CYP7A1. However, the role of the VDR in regulation of CYP7A1 has been controversial. Studies in the rat (Chow et al., 2009) and human primary hepatocytes and HepG2 cells suggested that the VDR inhibits CYP7A1 following treatment with 1,25(OH)2D3 (Han and Chiang, 2009; Han et al., 2010).

In mice, treatment with 1α(OH)D3, a vitamin D prodrug, resulted in upregulation of Cyp7a1 mRNA expression (Nishida et al., 2009). Meanwhile, treatment with a high dose of 1,25(OH)2D3 was shown to inhibit murine Cyp7a1 (Schmidt et al., 2010). The role of the VDR in cholesterol synthesis by regulation of HMGCR is also unclear. In vitro studies showed that treatment with vitamin D3 and

25(OH)D3 inhibited HMGCR expression, but 1,25(OH)2D3 treatment either inhibited or induced HMGCR depending on concentration and cell line (Gupta et al., 1989; Li et al., 2016). Considering all of these observations, the relationship between the VDR and cholesterol clearly requires further exploration.

As the VDR is found throughout the body and appears to play essential roles in human health and disease, maintaining sufficient vitamin D levels is critical for the VDR to properly exert its functions. Unfortunately, vitamin D deficiency is highly prevalent worldwide and has been associated with a number of diseases, including hyperparathyroidism, inflammation, hypertension, diabetes, cancer, and cardiovascular disease (Valdivielso et al., 2009). Interestingly, vitamin D deficiency and

hypercholesterolemia have been implicated in a number of clinical studies. However, intervention studies have provided mixed results (Ponda et al., 2012; Salehpour et al., 2012) and very few mechanistic evidence exists to link the two. Therefore, conclusive evidence that could correlate vitamin D status with cholesterol is of particular interest. Our laboratory has recently identified the

VDR as a regulator of CYP7A1, where 1,25(OH)2D3-liganded Vdr could downregulate Shp to increase Cyp7a1 and lower cholesterol levels in mice (Chow et al., 2014). Meanwhile, in rats fed a vitamin D-deficient diet, Vdr activity was reduced leading to upregulation of Srebp-2 and Hmgcr expression, thereby increasing cholesterol synthesis and contributing to the elevated serum cholesterol (Li et al., 2016). Hence, the relationship between vitamin D status, VDR expression, and the expression of the genes involved in cholesterol synthesis and metabolism require further exploration.

In this dissertation, we sought the direct mechanism and correlations between cholesterol, VDR, vitamin D status, and the expression of cholesterol-regulating genes. We first demonstrated that exogenous administration of 1,25(OH)2D3 could rapidly equilibrate into target tissues to activate the VDR and modulate transcription of VDR target genes. The VDR was also found to regulate cholesterol in a mechanism independent of the FXR. Since the Vdr was directly shown to downregulate Shp to increase Cyp7a1 for cholesterol lowering, we tested the corollary that vitamin D deficiency would decrease Vdr activity to upregulate Shp to decrease Cyp7a1 and thus increase cholesterol levels. Finally, since the VDR was identified as a potential therapeutic target for cholesterol lowering, we evaluated the PK and PD effects of 1,25(OH)2D3 treatment in vivo by developing a PKPD model and further extended this concept to a PBPK/PD model that could be used to adequately predict Cyp7a1 and cholesterol levels.

2.2 Hypotheses

1. The VDR is involved in cholesterol lowering in a mechanism that is independent of the FXR

2. Vitamin D deficiency raises cholesterol levels by altering the expression of cholesterol- regulating genes

3. The pharmacokinetic properties of 1,25(OH)2D3 are altered by its pharmacodynamics

2.3 Thesis outline

The objectives to test the hypotheses are outlined below:

1. To establish that 1,25(OH)2D3 can enter target tissues to exert direct effects on VDR target genes in mice in vivo (Chapter 3).

2. To identify that 1,25(OH)2D3 enters the liver freely and passively to activate the VDR and facilitate cholesterol lowering, in a mechanism that is independent of the FXR (Chapter 4).

3. To determine the effects of vitamin D deficiency on cholesterol and expression of cholesterol-regulating genes in mice in vivo (Chapter 5).

4. To assess the cholesterol lowering potencies of vitamin D analogs in mice fed a high fat/high cholesterol diet (Chapter 6).

5. To build a PKPD model to examine how changes in the synthesis (Cyp27b1) and degradation

(Cyp24a1) enzymes of 1,25(OH)2D3 alter its PK, and to extend this model to relate Cyp7a1 and cholesterol in the liver (Chapter 7).

Chapter 3

Temporal Gene Changes in Tissue 1α,25-Dihydroxyvitamin D3, Vitamin D Receptor Target Genes, and Calcium and PTH Levels

After 1,25(OH)2D3 Treatment in Mice

Edwin C.Y. Chow1, Holly P. Quach1, Reinhold Vieth2, and K. Sandy Pang1

1Department of Pharmaceutical Sciences, Leslie Dan Faculty of Pharmacy, University of Toronto, Toronto, Ontario, Canada

2Departments of Nutritional Sciences and Laboratory Medicine and Pathobiology, University of Toronto, Toronto, Ontario, Canada

Am J Physiol Endocrinol Metab 2013; 304:E977-E989.

3.1 Abstract

The vitamin D receptor (VDR) maintains a balance of plasma calcium and 1α,25- dihydroxyvitamin D3 [1,25(OH)2D3], its natural active ligand, by directly regulating the calcium ion channel (TRPV6) and degradation enzyme (CYP24A1), and indirectly regulating the parathyroid hormone (PTH) for feedback regulation of the synthetic enzyme CYP27B1. Studies that examined the intricate relationships between plasma and tissue 1,25(OH)2D3 levels and changes in VDR target genes and plasma calcium and PTH are virtually nonexistent. In this study,

we investigated temporal correlations between tissue 1,25(OH)2D3 concentrations and VDR target genes in ileum and kidney and plasma calcium and PTH concentrations in response to

1,25(OH)2D3 treatment in mice (2.5 µg/kg i.p., singly or q2d x4]. After a single i.p. dose, plasma

1,25(OH)2D3 peaked at ~0.5 h and then decayed biexponentially, falling below basal levels after 24 h and then returning to baseline after 8 days. Upon repetitive i.p. dosing, plasma, ileal, renal, and bone 1,25(OH)2D3 concentrations rose and decayed in unison. Temporal profiles showed increased expressions of ileal Cyp24a1 and renal Cyp24a1, Mdr1/P-gp, and VDR but decreased renal Cyp27b1 mRNA after a time delay in VDR activation. Increased plasma calcium and attenuated PTH levels and increased ileal and renal Trpv6 expression paralleled the changes in tissue 1,25(OH)2D3 concentrations. Gene changes in the kidney were more sustained than those in intestine, but the magnitudes of change for Cyp24a1 and Trpv6 were lower than those in intestine.

The data revealed that 1,25(OH)2D3 equilibrates with tissues rapidly, and VDR target genes

respond quickly to exogenously administered 1,25(OH)2D3.

3.2 Introduction

1α,25-Dihydroxyvitamin D3 [1,25(OH)2D3] is the active form of vitamin D and natural ligand of the vitamin D receptor (VDR), a member of the steroid/thyroid hormone nuclear receptor superfamily (Mangelsdorf et al., 1995). Vitamin D is formed from 7-dehydrocholesterol in skin upon exposure to the ultraviolet rays of the sun (270-300 nm) and is hydroxylated by the cytochrome P-450s (CYP2R1 and CYP27A1) in liver to form 25-hydroxyvitamin D3 [25(OH)D3]. This inactive metabolite is transported by the vitamin D binding protein (DBP) and taken up by

the endocytic receptor megalin into renal proximal tubular cells (Nykjaer et al., 1999; Takemoto et al., 2003) for activation by the 1α-hydroxylase (CYP27B1) to form the active metabolite,

1,25(OH)2D3 (Jones et al., 1998; Takeyama and Kato, 2011).

Upon binding of 1,25(OH)2D3 to VDR, the complex undergoes a conformational change and translocates to the nucleus to heterodimerize with the retinoid X receptor (RXR) (Lemon and Freedman, 1996; Bettoun et al., 2003), followed by recruitment of coactivators before binding to vitamin D response elements (VDREs) in promoter regions of VDR-responsive genes to initiate

gene transcription (Dusso et al., 2005). One of the physiological roles of 1,25(OH)2D3 is to increase plasma calcium levels through transactivation of the calcium ion channels [transient receptor potential cation channel subfamily V members 5 and 6 (TRPV5 and TRPV6)] in the kidney and intestine (Suzuki et al., 2008). It is known that calcium is maintained by the concerted actions in not only the epithelia of the kidney and intestine, but also bone, where turnover is a continuous process involving both resorption of existing bone and deposition of new bone, processes that are stimulated by actions of 1,25(OH)2D3 and the parathyroid hormone, PTH (Jones et al., 1998; Hoenderop et al., 2005).

The level of 1,25(OH)2D3 in plasma is tightly controlled by two major cytochrome enzymes in the

kidney, CYP27B1 or the 1α-hydroxylase, which converts 25(OH)D3 to 1,25(OH)2D3, and

CYP24A1, the catabolic enzyme that degrades 25(OH)D3 and 1,25(OH)2D3 to 24,25(OH)2D3 and

1,24,25(OH)3D3, respectively (Feldman et al., 2005). CYP24A1 is present in tissues that express VDR (Armbrecht and Boltz, 1991) and serves as a biomarker for VDR activation. When

1,25(OH)2D3 in plasma is high, CYP24A1 becomes highly induced in the kidney to increase catabolism (Meyer et al., 2007). When the circulating calcium concentration is low, the parathyroid gland responds by stimulating the production of PTH to downregulate CYP24A1 mRNA stability

in kidney, reducing 1,25(OH)2D3 degradation to result in higher plasma levels of 1,25(OH)2D3 (Zierold et al., 2001; Zierold et al., 2003). When plasma calcium level is high, activation of the calcium sensing receptor (CaSR) leads to reduction of PTH (Jones et al., 1998; Feldman et al.,

2005). In the intestine, CYP24A1 level is regulated by 1,25(OH)2D3 and not PTH (Shinki et al., 1992), suggesting that induction of CYP24A1 in the intestine is more of an acute response to VDR compared to the kidney (Henry, 2001).

Within the past decade, the VDR has been implicated to play an important role in the regulation of drug enzymes and transporters. VDR-responsive drug-metabolizing enzymes include the cytochrome P-450s [human CYP3A4, CYP2B6, and CYP2C9 (Thummel et al., 2001; Drocourt et al., 2002), and rodent Cyp3a1, Cyp3a9, and Cyp24a1 (Chow et al., 2009; Chow et al., 2010)] and sulfotransferase 2A1 (SULT2A1) (Echchgadda et al., 2004). VDR-responsive transporters include the rat apical sodium dependent bile acid transporter (Asbt) (Chen et al., 2006), human organic anion-transporting polypeptide (OATP1A2) (Eloranta et al., 2012), multidrug resistance protein-1 or P-glycoprotein (MDR1/P-gp) (Saeki et al., 2008), and the human and rodent multidrug resistance-associated proteins (MRP2/Mrp2, Mrp3, MRP4/Mrp4) both in vitro and in vivo (Fan et al., 2009; Chow et al., 2010). Our laboratory has shown that VDR transactivates P-gp in brain microvessel endothelia (Durk et al., 2012) in vitro and P-gp in murine kidney and brain but not ileum and liver in vivo, leading to hastened efflux of digoxin in the brain and kidney (Chow et al., 2011a).

Clinically, the concentration of 25(OH)D3 in plasma (nM range) is used as the biomarker for vitamin D status, though this value does not reflect the concentration of its active metabolite,

1,25(OH)2D3 (pM range) formed via CYP27B1, the rate-limiting enzyme (Dusso et al., 2005; Wang et al., 2008). Thus, there is the need to define precisely the pharmacological effects and interplay between 1,25(OH)2D3 and the respective VDR target genes. A temporal study was thus undertaken to examine tissue levels of 1,25(OH)2D3 on changes in VDR target genes in mice in a time-dependent manner after repeated administration of small doses of 1,25(OH)2D3 (2.5 µg/kg i.p. q2d x 4). We aimed to test the hypotheses that 1,25(OH)2D3 enters and equilibrates with tissues readily and that effects of VDR-responsive genes are delayed after entry of 1,25(OH)2D3 in the tissue, since signal transduction is a multistep process involving multiple organs and feedback mechanisms.

3.3 Materials and methods

3.3.1 Materials

1,25(OH)2D3 in powder form was purchased from Sigma-Aldrich Canada (Mississauga, ON). Antibodies to Cyp24 (H-87) and Villin (C-19) were purchased from Santa Cruz Biotechnology (Santa Cruz, CA), anti-Gapdh (cat. no. ab8245) and anti-Trpv6 (cat. no. ab63084) from Abcam (Cambridge, MA), and P-gp from ID Labs (London, ON). All other reagents were purchased from Sigma-Aldrich Canada (Mississauga, ON) and Fisher Scientific (Mississauga, ON). The enzyme-

immunoassay (EIA) kit (cat. no. AC-62F1) for 1,25(OH)2D3 measurement was manufactured by Immunodiagnostics Systems (IDS) Inc. (Scottsdale, AZ) and purchased from Inter Medico (Markham, ON). The mouse PTH 1-84 ELISA kit, manufactured by Immutopics International (San Clemente, CA), was obtained via Joldon Diagnostics (Burlington, ON).

3.3.2 Pharmacokinetic study of 1,25(OH)2D3 in mice

Anhydrous ethanol was used to dissolve the 1,25(OH)2D3 powder, and the resulting concentration was measured spectrophotometrically at 265 nm (UV-1700, Shimadzu Scientific Instruments) before dilution with sterile corn oil. For in vivo pharmacokinetic studies in male, C57BL/6 mice

(8-weeks old), doses of either 0 or 2.5 µg/kg (or 0.05 µg/mouse or 120 pmol) 1,25(OH)2D3, dissolved in sterile corn oil (5 μl/g), were given i.p. on Days 0, 2, 4, and 8 at 9 am. The dose was based on a series of published reports, with doses ranging from 0.1 to 5 µg/mouse or 0.25 to 5 µg/kg for i.p. or oral administration, daily or every other day in mice (Yu et al., 1998; Hershberger et al., 2001; Prudencio et al., 2001; Ahmed et al., 2002; Muindi et al., 2004; Albert et al., 2007;

Muindi et al., 2010; Swami et al., 2011). In humans, doses of 1,25(OH)2D3 (0.5 µg/kg or 38 µg) furnished a similar clearance and exposure with minimal toxicity when used intermittently (Brandi et al., 2002; Muindi et al., 2002). Moreover, preliminary studies had shown that the chosen dosing regimen in mice elicited the desired pharmacologic effect with only minor elevation of plasma calcium.

One mouse was sacrificed at each blood sampling point for the treatment groups, and tissues were harvested at 0, 0.08, 0.25, 0.5, 1, 3, 6, 9, 12, 24, 48, 96, 192, and 360 h after the first dose on Day 0, and at 0, 0.5, 1, 3, 6, 9, 12, 24, and 48 h, after subsequent doses [in accordance with animal

protocols approved by the University of Toronto (ON, Canada)]. For the control group (treated with corn oil only), however, sampling was conducted at 0, 3, 6, and 12 h on Day 0, and at 0 h on Days 2, 4, 6, 8, and 14. The mouse was rendered unconscious in a carbon dioxide chamber before blood collection by cardiac puncture using a 1-ml syringe-23G 3/4 inch needle set that was prerinsed with heparin (1000 IU/ml). Plasma was obtained by centrifugation of blood at 3,000 rpm for 10 min. After flushing of ice-cold saline through the lower vena cava for the removal of blood, the kidneys were harvested, weighed, reduced to small pieces, snap-frozen in liquid nitrogen, and stored at -80°C for future analyses (Chow et al., 2011a). The small intestinal segments: the duodenum (spanning from the pyloric ring to the ligament of Treitz), proximal jejunum (6 cm after the ligament of Treitz), and ileum (6 cm proximal to the ileocecal junction) were removed as described (Chow et al., 2009). After flushing of the segments with cold 1 mM phenylmethylsulfonyl fluoride (PMSF) saline solution, the segments were everted and scraped, and the scrapings were snap-frozen in liquid nitrogen and stored at -80°C. The colon was also removed, flushed with cold 1 mM PMSF saline solution, blot-dried, snap-frozen in liquid nitrogen and stored at -80°C. For isolation of the femur (bone), the skin, muscle, tendon, and fat around the leg were removed and/or scraped off with a razor blade and then snap-frozen in liquid nitrogen and stored at -80°C until analyses.

3.3.3 Plasma calcium and phosphorus analyses and PTH assay

Calcium and phosphorus measurements in plasma were determined by inductively coupled plasma atomic emission spectroscopy (ICP-AES; Optima 3000 DV, Perkin Elmer) as previously described (Chow et al., 2011b). Plasma was diluted 350-fold with 1% nitric acid before each measurement. Calcium was measured at 317.9 nm and 315.9 nm and phosphorus at 213.6 nm and 214.9 nm. PTH levels in 20 µl plasma were assayed by ELISA according to the manufacturer’s protocol.

3.3.4 Tissue 1,25(OH)2D3 extraction and 1,25(OH)2D3 EIA for plasma and tissue samples

The tissue extraction procedure for lipids was similar to Bligh and Dyer (1959), with modifications described by Wagner et al. (2012). Weighed kidney and scraped intestinal enterocyte samples were added to double-distilled water (wt/vol) to a final volume of 1 ml. The sample was homogenized

with 3.75 ml of methylene chloride and methanol (1:2 vol/vol). The weighed femur was added to double-distilled water (wt/vol up to 1 ml) and 3.75 ml of a mixture of methylene chloride and methanol (1:2 vol/vol) before the bone mixture was crushed using a mortar and pestle to obtain a homogenate. The final homogenate was mixed with 1.25 ml methylene chloride and vortexed for 1 min, and then 1.25 ml double-distilled water was added and further mixed for another minute before centrifugation at 3,000 rpm for 20 min at room temperature. The methylene chloride (bottom phase) was collected using a glass pasteur pipette. Extraction of the homogenate was repeated upon addition of 1.25 ml of methylene chloride. The harvested methylene chloride extract was pooled, dried under nitrogen gas, and reconstituted with 0.3 ml of charcoal-stripped human

serum (Wagner et al., 2012). The concentration of 1,25(OH)2D3 in mouse plasma or tissue was

determined using the 1,25(OH)2D3 EIA kit following the manufacturer’s protocol. Plasma and tissue samples, when out of the calibration range, were diluted with charcoal-stripped human serum before analysis. Delipidated solutions of tissue samples were spun at 12,000 g for 10 min through a 0.2 µm Nanosep MF Centrifugal Device (Pall Life Sciences, Ville St. Laurent, QC) before addition to the immunocapsules for EIA.

3.3.5 Preparation of subcellular protein fractions of kidney and intestinal tissues

For preparation of the crude membrane fraction for the assay of Cyp24 or P-gp protein, kidney, colon tissue, or scraped enterocytes were homogenized in the crude membrane homogenizing buffer (250 mM sucrose, 10 mM HEPES, and 10 mM Trizma base, pH 7.4) containing 1% protease inhibitor cocktail (Chow et al., 2011a). This homogenate was used for Western blotting to measure total VDR protein expression. The homogenate was then centrifuged at 3,000 g for 10 min at 4°C, and the resulting pellet containing the mitochondrial fraction was resuspended in a buffer (in mM:

60 KCl, 15 NaCl, 5 MgCl2•6H2O, 0.1 EGTA, 300 sucrose, 0.5 DTT, 0.1 PMSF, and 15 Trizma HCl, pH 7.4) containing 1% protease inhibitor cocktail for Western blotting to measure Cyp24 protein expression. The supernatant was spun at 20,000 g for 60 min at 4°C, and the resulting pellet was resuspended in the resuspension buffer, as previously described (Chow et al., 2009) for Western blotting to measure P-gp and Trpv6 protein expression. Protein concentrations of the samples were assayed by the Lowry method (Lowry et al., 1951) using bovine serum albumin as the standard. Samples were then stored at -80°C until Western blot analyses.

3.3.6 Western blotting

Total protein samples (15-50 µg) were separated by 7.5-10% SDS-polyacrylamide gels at 100 V according to Chow et al. (2011a). After separation and transfer of proteins onto a nitrocellulose membrane (Amersham Biosciences, Piscataway, NJ), the membrane was blocked with 5% (wt/vol) skim milk in Tris-buffered saline (pH 7.4) and 0.1% Tween 20 (TBS-T; Sigma-Aldrich, Mississauga, ON) for 1 h at room temperature and washed once with 0.1% TBS-T, followed by incubation with primary antibody solution in 2% skim milk in 0.1% TBS-T overnight at 4°C. On the next day, the membrane was washed with 0.1% TBS-T and then incubated with secondary antibody in 2% skim milk in 0.1% TBS-T for 2 h at room temperature and again washed with 0.1% TBS-T. Bands were visualized using chemiluminescence reagents (Amersham Biosciences) and quantified by scanning densitometry (NIH Image software; http://rsb.info.nih.gov/nih-image/). Band intensity of the target protein was normalized to the protein band intensity of Villin (95 kDa) for the intestinal samples and Gapdh (36 kDa) for the kidney samples. For comparison of intestinal and renal protein expression, both samples were normalized to that of Gadph.

3.3.7 Quantitative real-time PCR

Total RNA from kidney tissue and scraped ileal enterocytes was extracted with the TRIzol extraction method (Sigma-Aldrich) according to the manufacturer’s protocol, with modifications (Chow et al., 2011a). About 1.5 µg of cDNA was immediately synthesized from the RNA samples, using the High Capacity cDNA Reverse Transcription Kit (Life Technologies Canada, Burlington, ON), and qPCR was performed with SYBR Green detection system as described (Chow et al., 2011a). Information on primer sequences is listed in Table 3-1. All target mRNA data were normalized to Cyclophilin mRNA for the kidney samples and Villin mRNA for the intestinal samples, and the calculation of relative change in gene expression was performed as previously described (Chow et al., 2009).

Table 3-1. Mouse primer sets for quantitative real-time PCR

Gene Bank Forward (5’→ 3’ Sequence) Reverse (5’→ 3’ Sequence) Number Cyp24a1 NM_009996 CTGCCCCATTGACAAAAGGC CTCACCGTCGGTCATCAGC Mdr1 NM_011076 TACGACCCCATGGCTGGATC GGTAGCGAGTCGATGAACTG Trpv6 NM_022413 ATCGATGGCCCTGCCAACT CAGAGTAGAGGCCATCTTGTTGCTG VDR NM_009504 GAGGTGTCTGAAGCCTGGAG ACCTGCTTTCCTGGGTAGGT Cyclophilin X58990 GGAGATGGCACAGGAGGAA GCCCGTAGTGCTTCAGCTT Villin NM_009509 TCCTGGCTATCCACAAGACC CTCTCGTTGCCTTGAACCTC

3.4 Results

3.4.1 Similar plasma and tissue (ileum, kidney, and bone) decay of

1,25(OH)2D3 after single and multiple dosing of 1,25(OH)2D3 to mice

The baseline plasma concentration of 1,25(OH)2D3 for vehicle-treated C57BL/6 mice, estimated as the mean of the determinations for the experimental duration, was 212 ± 29 pM (fmol/ml), a

value similar to the endogenous plasma concentration of 1,25(OH)2D3 in the rat (Vieth et al.,

1990a) but higher than that in man (Brandi et al., 2002). Basal levels of 1,25(OH)2D3 in the kidney (70.5 ± 10.4 pmol/kg tissue), intestine (93.5 ± 7.2 pmol/kg tissue), and bone (36.5 ± 33.8 pmol/kg tissue) were significantly lower than those in plasma (Fig. 3-1A); the tissue/plasma concentration ratios were 0.35, 0.4, and 0.17 for the kidney, ileum, and bone, respectively (Fig. 3-1B). The less-

than-unity tissue partitioning ratio may be explained by the high plasma binding of 1,25(OH)2D3 relative to those in tissues.

(A) (B)

300 0.6

250 0.5

200 0.4

150 0.3 Concentration 3

D 2 * * 100 * 0.2 (pM or pmol/kg)(pM or 50 0.1 Tissue to Plasma Ratio

1,25(OH) 0 0.00 Plasma Ileum Kidney Bone Ileum Kidney Bone

Figure 3-1. Basal levels of 1,25(OH)2D3 in plasma, kidney, ileum, and bone (A), and tissue/plasma concentration ratios (B) in untreated C57BL/6 control mice. Data represent means ± SEM (n = 3-4). *P < 0.05, basal tissue vs. plasma 1,25(OH)2D3 levels (Mann-Whitney U-test).

After a single i.p. dose of 50 ng/mouse (120 pmol/mouse or 2.5 µg/kg), 1,25(OH)2D3 was rapidly

absorbed (tmax or time for maximum concentration of ≤0.5 h), yielding a peak plasma concentration

(Cmax) of 44 nM, followed by an apparent, biexponential decay profile (Fig. 3-2A) with a half-life

of ~6.86 h (Table 3-2). Distinctively, the plasma concentration of 1,25(OH)2D3 fell below the baseline by the end of the Day 0, reached a nadir by Day 4, and returned to baseline levels by the

Day 8 (Fig. 3-2A). Subsequent to i.p. dosing of 1,25(OH)2D3, tissue levels in the kidney, ileum, and bone peaked at similar times [6, 4.5, and 0.065 nmol/kg (or pmol/g) tissue, respectively] and decayed in parallel fashion to that of plasma (Fig. 3-2A).

Table 3-2. Non-compartmental estimates for 1,25(OH)2D3, after repeated doses of 2.5 µg/kg q2d x 4 i.p. to mice Dose 1 Dose 2 Dose 3 Dose 4 Dose (pmol) 120 120 120 120 Terminal decay 0.10 0.09 0.10 0.11 constant, β (h-1)a b t1/2,β (h) 6.86 7.48 6.79 6.31 c AUC0→48 (nM*h) 179 131 170 117 a Terminal decay constant (β) was estimated from the negative slope of ln(concentration) vs. time data from averaged data between 12-48 h after i.p. dosing. b Terminal half life (t1/2,β) was calculated as 0.693/β. c Area under the curve [AUC(0→48)] between 0 to 48 h was estimated by the trapezoidal rule.

Following multiple 1,25(OH)2D3 i.p. dosing, patterns of decay for plasma 1,25(OH)2D3 were similar after each injection, wherein the 1,25(OH)2D3 concentration peaked (averages of 44, 22, 43, and 29 nM) between 0.5 and 1 h after each of the injections given. Levels again fell below the basal level by the end of 24 h postinjection (Fig. 3-2B). The plasma 1,25(OH)2D3 concentration before the next injection at the nadir was much lower than of the basal level, and the same pattern

persisted throughout the dosing regimen. The renal, ileal, and bone 1,25(OH)2D3 tissue concentrations rose in union to those in plasma, reaching a peak concentration of 6.1, 1.8, 5.0, and 5.6 nmol/kg tissue in kidney, 4.5, 1.9, 1.9, and 2.8 nmol/kg tissue in ileum, and 0.065, 0.505, 7.65, and 1.8 nmol/kg tissue in bone, respectively, between 0.5 and 3 h after each of the four injections.

These data confirm that 1,25(OH)2D3 is able to equilibrate readily between plasma and tissue.

Multiple dosing of 1,25(OH)2D3 resulted in a terminal or beta half-life of ~6.3 to 7.5 h (Fig. 3-2B), and no trend was discernable upon repeated dosing (Table 3-2). Overall, the decay pattern of

1,25(OH)2D3, based on the total 1,25(OH)2D3 (exogenous + basal) concentration after the

administered 1,25(OH)2D3, and the exposure (AUC0-48) were similar after each injection (Fig. 3-

2B, Table 3-2). There was little change in the pharmacokinetics of 1,25(OH)2D3 upon repeated dosing, since change in the enzyme for catabolism, Cyp24a1, was maximal after a single dose (see results below in intestine and kidney). The half-life is similar to that observed from other i.p. studies in the mouse (7.6 h) (Muindi et al., 2004).

Figure 3-2. Plasma and tissue 1,25(OH)2D3 (ileum, kidney, and bone) concentration-time profiles from a single dose (A) or multiple doses (B) (Days 0, 2, 4, and 6) of 2.5 µg/kg i.p. 1,25(OH)2D3 q2d x 4 to mice. Data for vehicle-treated mice (control) were averaged and are denoted as open circles interconnected by solid line (n = 2-4). For treated mice, individual 1,25(OH)2D3 datum is shown (filled circle, one mouse per sample); averaged values are joined by the dashed line (n = 2-4).

3.4.2 Intestinal distribution and effects of 1,25(OH)2D3 on intestinal and colon VDR, Cyp24a1, and Trpv6 mRNA expression

The distribution of basal VDR mRNA expression was found to be higher for the duodenum than jejunum (50% of duodenum) and ileum (46% of duodenum) but was highest in the colon (1.5-fold of duodenum) (Fig. 3-3A, left). Basal Cyp24a1 mRNA expression was evenly distributed in the small intestine but was highest in colon (42-fold of duodenum) (Fig. 3-3B, left). The basal mRNA expression of Trpv6 was highest in the duodenum, followed by the colon (46% of duodenum), and

was negligible in the jejunum and ileum (<1%) (Fig. 3-3C, left). At 3 h post-1,25(OH)2D3 injection, there was no major change in VDR mRNA expression for all intestinal segments and the colon (Fig. 3-3A, left). By contrast, Cyp24a1 and Trpv6 mRNA expressions were elevated >900- fold and >7-fold in the duodenal and ileal segments, respectively, although not for the jejunum due to sample variation (Figs. 3-3, B and C, left). The lack of Cyp24a1 mRNA induction and small

Trpv6 mRNA change in colon (Figs. 3-3, B and C, left) with 1,25(OH)2D3 treatment agrees with

the possibility that very low amounts of 1,25(OH)2D3 are available to enter into colon due to the route of administration (i.p.).

Focusing on the ileum, where induction of Asbt (an important VDR target gene to transport bile acids) was previously found to occur in the rat (Chen et al., 2006; Chow et al., 2009), single or repetitive treatment of 1,25(OH)2D3 to mice elicited only minimal changes in ileal VDR mRNA

expression (Fig. 3-3A, middle and right), though the temporal ileal 1,25(OH)2D3 concentrations rose in unison to that in plasma and remained mostly above basal levels during each injection interval (Fig. 3-2). By contrast, ileal Cyp24a1 mRNA was induced significantly between 3 and 9 h after dosing (Fig. 3-3B, middle and right), with patterns similar to the temporal changes of

1,25(OH)2D3 in ileum (Fig. 3-2). The induction of Cyp24a1 mRNA was dramatically increased (> 500-fold) after the first dose (Fig. 3-3B, middle), but the inductions were slightly lessened (300- 400-fold) for the 2nd, 3rd, and 4th doses, with mRNA levels rapidly returning to baseline at 24 h

after dosing (Fig. 3-3B, right). A single administration of 1,25(OH)2D3 resulted in a 30-fold maximal increase in ileal Trpv6 mRNA at 9 h (Fig. 3-3C, middle). However, multiple dosing of

1,25(OH)2D3 greatly magnified the increase of ileal Trpv6 to 200- to 600-fold (Fig. 3-3C), with higher changes observed at the 3rd and 4th doses (Fig. 3-3C, middle and right).

Figure 3-3. Intestinal distribution of mRNA and effect of 1,25(OH)2D3 (left) on vitamin D receptor (VDR; A), degradation enzyme (Cyp24a1; B), and calcium channel (Trpv6; C) mRNA expression in duodenum, jejunum, ileum, and colon at 3 h post-2.5 µg/kg i.p. 1,25(OH)2D3 injection; temporal changes for ileal mRNA of VDR (A), Cyp24a1 (B), and Trpv6 (C) after a single dose (middle) or multiple doses (right) of 1,25(OH)2D3. Data at left represents mean ± SEM (n = 3 or 4). In left, †P < 0.05 between basal duodenal control vs. basal control of other intestinal segments; *P < 0.05, basal control vs. 1,25(OH)2D3–treated group (Mann-Whitney U-test). In middle and right, data for vehicle-treated mice (control) were averaged and are denoted as open circles interconnected by solid line (n = 2-4). For treated mice, individual 1,25(OH)2D3 datum is shown (filled circle); averaged values are joined by dashed line (n = 2-4).

3.4.3 Induction of renal VDR, Cyp24a1, and Trpv6 and downregulation of Cyp27b1 mRNA were time and concentration dependent

Unlike the modest change of VDR mRNA observed for the intestine, renal VDR mRNA rose 2- fold between 3 and 12 h after the single 1,25(OH)2D3 dose, and this increase was sustained for the entire week before levels returned to baseline. With multiple dosing, renal VDR mRNA levels for subsequent doses were further increased to 2.5- then 4-fold higher over basal levels (Fig. 3-4A).

After a single 1,25(OH)2D3 dose, renal Cyp24a1 mRNA expression was increased 77-fold and peaked at ~9 h, and levels remained high and above baseline for 8 days, whereas multiple dosing

of 1,25(OH)2D3 induced and maintained Cyp24a1 mRNA over 40-fold above basal levels (Fig. 3-

4B). Injection of a single 1,25(OH)2D3 dose resulted in a 5-fold maximal increase in renal Trpv6 mRNA at 9 h, whereas multiple dosing greatly magnified the increase of renal Trpv6, with higher changes observed for Trpv6 mRNA at the 3rd and 4th doses of ~10- to 12-fold above basal levels (Fig. 3-4C). For renal Cyp27b1 mRNA, there was an immediate rise above basal level at 0.5 h following the first injection, but this change was followed by a rapid decrease to 34% of basal level at 3 h (Fig. 3-4D); these levels were maintained below basal levels (8-30%) before returning back to baseline on the 8th day, and the average renal Cyp27b1 mRNA level was ~3-30% of basal levels during repetitive dosing (Fig. 3-4D), suggesting that endogenous 1,25(OH)2D3 synthesis was likely reduced.

Figure 3-4. Temporal changes in renal mRNA expression of VDR (A), Cyp24a1 (B), Trpv6 (C), and synthetic enzyme, Cyp27b1 (D), after a single dose or multiple doses of 1,25(OH)2D3 to mice. Data for vehicle-treated mice (control) were averaged and are denoted as open circles interconnected by solid line (n = 2-4). For treated mice, individual 1,25(OH)2D3 datum is shown (filled circle); averaged values are joined by dashed line (n = 2-4).

3.4.4 Induction of renal Mdr1 mRNA and P-gp protein by 1,25(OH)2D3

Similarly, renal Mdr1 mRNA showed a relatively small increase (1.5- to 2-fold) before returning back to basal level 6 h after the single administration (Fig. 4-5A). However, upon repetitive dosing, a disproportionate and sustained increase (> 10-fold) was observed (Fig. 4-5B). Renal P-gp protein

levels resulting from 1,25(OH)2D3 repetitive treatment rose on average 5-fold above basal levels (Fig. 4-5C).

Figure 3-5. Temporal changes in renal mRNA and protein expression of multidrug resistance protein-1 or P-glycoprotein (Mdr1/P-gp; 170 kDa) from a single dose (A) or multiple doses (B and C) of 1,25(OH)2D3 to mice. Data for vehicle-treated mice (control) were averaged and are denoted as open circles and joined by solid line (n = 2-4). For treated mice, individual 1,25(OH)2D3 datum is shown (filled circle); averaged values are joined by dashed line (n = 2-4).

3.4.5 Temporal changes in intestinal and renal VDR, Cyp24, and Trpv6 protein expression vs. plasma calcium and PTH levels in single and multiple doses of 1,25(OH)2D3

Basal protein level of VDR was higher in the duodenum than ileum (30% of duodenum) but was highest in colon (4-fold; Fig. 3-6A), whereas VDR protein was similar between the duodenum and kidney (Fig. 3-6B). Changes in protein expression of VDR were different compared with its mRNA expression (Fig. 3-3A). When examined, VDR protein in the various segments showed

differential temporal changes with 1,25(OH)2D3 treatment for the duodenum, ileum, and colon.

nd rd VDR protein in duodenum was elevated after the 2 and 3 1,25(OH)2D3 injections, whereas VDR protein level in the colon was increased after the 3rd and 4th injections (Fig. 3-6C). There were higher protein changes but a lack of change in ileal VDR mRNA with treatment (Fig. 3-3A), and

this could be explained by the action of 1,25(OH)2D3 in increasing the half-life of VDR protein (Karnauskas et al., 2005). Renal VDR protein, similar to VDR mRNA, rose rapidly after the 1st dose, and was sustained for the 2nd and 4th doses (Fig. 3-6C). Furthermore, multiple administration

of 1,25(OH)2D3 increased Cyp24 protein in ileum steadily at 1.5-fold above basal level (Fig. 3- 7A), whereas the changes for renal Cyp24 protein were considerably higher (10-fold on average) than in ileum (Fig. 3-7B).

Figure 3-6. Distribution (A and B) and temporal changes (C) of VDR protein (54 kDa) in duodenum, jejunum, ileum, colon, and kidney after multiple doses of 1,25(OH)2D3. (A): comparison of VDR protein in different intestinal segments was normalized to Villin. (B): comparison of VDR protein between duodenum and kidney was normalized to Gapdh. Data in (A) and (B) represent means ± SEM (n = 3-4). (A): †P < 0.05, basal duodenal control vs. basal control of other intestinal segments (Mann-Whitney U-test). (C): data for vehicle-treated mice (control) were averaged and are denoted as open circles interconnected by solid line (n = 2-4). For treated mice, individual 1,25(OH)2D3 datum are shown (filled circle); averaged values are joined by dashed line (n = 2-4).

Figure 3-7. Temporal changes in ileal (A) and renal (B) relative protein expression of Cyp24 (55 kDa) after multiple doses of 1,25(OH)2D3. Data for vehicle-treated mice (control) were averaged and are denoted as open circles joined by solid line (n = 2-4). For treated mice, individual 1,25(OH)2D3 datum is shown (filled circle); averaged values are joined by dashed line (n = 2-4).

We then examined basal levels of Trvp6 protein in the duodenum, ileum, and colon and found the rank order of duodenum > colon (72% of duodenum) > ileum (25% of duodenum) (Fig. 3-8A). The basal level of Trpv6 protein of the kidney was only 43% of that in duodenum (Fig. 3-8B).

With 1,25(OH)2D3 treatment, Trpv6 protein levels were increased 1.5-fold on average throughout the treatment period for the duodenum and ileum after the 3rd and 4th doses, whereas levels in the colon were increased after the 4th dose (Fig. 3-8C). These changes in protein are consistent with a high Trpv6 mRNA induction in ileum at the same period (Fig. 3-3C).

Furthermore, levels of plasma calcium were correlated to changes in Trpv6 mRNA (Fig. 3-3C) and protein (Fig. 3-8C). There was virtually no discernable change in calcium levels after a single dose, whereas cumulative changes were observed after the 2nd to 4th doses (Fig. 3-8D). These changes could be attributed to the relatively high and transient elevation of both the intestinal and renal Trpv6 mRNA, raising the plasma calcium concentrations by 10-40% during the successive dosing regimen (Fig. 3-8D). However, due to the 2.2-fold higher Trpv6 protein level in the duodenum compared with that in kidney (Fig. 3-8B), the intestine is likely a greater contributor to calcium absorption. This notion was supported by others as well (Xue and Fleet, 2009; Cui et al., 2012).

Figure 3-8. Distribution (A and B) and temporal changes (C) of Trpv6 (89 kDa) relative protein expression in duodenum, jejunum, ileum, colon, and kidney after multiple doses of 1,25(OH)2D3. (D): temporal changes of plasma calcium from a single dose or multiple doses of 1,25(OH)2D3 to mice. (A): comparison of Trpv6 protein in different intestinal segments was normalized to Villin. (B): comparison of Trpv6 protein between duodenum and kidney was normalized to Gapdh. Data for (A) and (B) represent means ± SEM (n = 3-4). (A): †P < 0.05, basal duodenal control vs. basal control of other intestinal segments (Mann-Whitney U-test). (B): #P < 0.05, basal duodenal control vs. basal kidney control (Mann-Whitney U-test). (C) and (D): data for vehicle-treated mice (control) were averaged and are denoted as open circles interconnected by solid line (n = 2-4). For treated mice, individual 1,25(OH)2D3 datum is shown (filled circle); averaged values are joined by dashed line (n = 2-4).

The mean basal plasma PTH level was ~68.8 ± 14.8 pg/ml in control mice (Fig. 3-9). After a single dose of 1,25(OH)2D3, mouse plasma PTH initially increased (to ~ 148 pg/ml) in the first 5 min postinjection, but levels then immediately dropped to 12-34 pg/ml between 6 and 48 h before

returning back to basal levels on the eighth day (Fig. 3-9A). Multiple dosing of 1,25(OH)2D3 to mice generally led to sustained decreased plasma PTH (between 0 and 30 pg/ml) throughout the course of treatment, except at two sampling time points (0.5 h after 3rd injection and 9 h after 4th injection; Fig. 3-9B), likely due to sampling variation and small sample size.

Figure 3-9. Temporal changes in plasma PTH concentration from a single dose (A) or multiple doses (B) of 1,25(OH)2D3 to mice. Data for vehicle-treated mice (control) were averaged (n = 11), assumed to be identical and are denoted as open circles joined by solid line. For treated mice, individual 1,25(OH)2D3 datum is shown (filled circle); averaged values are joined by dashed line (n = 1-3).

3.5 Discussion

Our efforts represent one of the first studies to examine plasma and tissue 1,25(OH)2D3 concentrations accompanying the exogenous 1,25(OH)2D3 administration to mice. Despite being tightly bound to the vitamin D binding protein (DBP) in the plasma, we observed rapid distribution of 1,25(OH)2D3 into tissues due to the lipophilic nature of the compound (Dusso et al., 2005).

1,25(OH)2D3 is able to enter and distribute into tissues rapidly, including the kidney, intestine, bone (Fig. 3-1), liver, and brain (data not shown) regardless of differences in VDR abundance. The

parallel patterns of rise and decay for 1,25(OH)2D3 suggest rapid entry, distribution, and

equilibrium between tissue and plasma. 1,25(OH)2D3 concentrations in the kidney and intestine

remained predominantly above basal levels during and after 1,25(OH)2D3 treatment (Fig. 3-2), implying that the present regimen for treatment of 1,25(OH)2D3 could result in sustained local pharmacological effects. Under basal conditions, the tissue-to-plasma partitioning ratios are for ileum 0.41 ± 0.12, for kidney 0.33 ± 0.07, and for bone 0.15 ± 0.16 (Fig. 3-1). Those in the liver (0.13 ± 0.04) and brain (0.007 ± 0.003) were also lower (data not shown).

Upon correlation of the plasma and tissue 1,25(OH)2D3 concentration-time profiles to changes of VDR target genes in mice, we found that maximal induction of VDR target genes such as Trpv6 and Cyp24a1 mRNA expression in intestine were similar after single vs. repeated dosing (Figs. 3- 3, B and C), with the peak occurring between 3 and 9 h postinjection, lagging behind the peak

1,25(OH)2D3 concentration in ileum (at ~0.5-1 h) (Fig. 3-2). This lag time is not unexpected, and is likely the result of the time required for translocation of the VDR into the nucleus for heterodimerization with the RXR to initiate transcription. Changes in renal VDR, Cyp24, and Mdr1 mRNA expression also showed the time lag but were more sustained after repeated dosing,

since the VDR level was elevated and was more responsive to 1,25(OH)2D3 treatment (Figs. 3-4 and 3-5). In addition, tissue ileal and kidney concentrations of 1,25(OH)2D3 at 3-9 h after each

1,25(OH)2D3 administration remained elevated above baseline values (Fig. 3-2), at which time maximal induction of renal and ileal Cyp24a1, VDR and Trpv6 mRNA expression was noted (Figs. 3-3 and 3-4).

Cyp24a1 is a major VDR-responsive gene (Chen and DeLuca, 1995; Itoh et al., 1995) that

metabolizes 1,25(OH)2D3 to 1,24,25-trihydroxyvitamin D3 and 25-hydroxyvitamin D3 to 24,25- dihydroxyvitamin D3 (Jones et al., 1998), and absence of Cyp24a1 in Cyp24 knockout mice

drastically reduces 1,25(OH)2D3 metabolism (Masuda et al., 2005). In this study, we found that induction of ileal Cyp24a1 mRNA (Fig. 3-3B) was much greater than that for renal Cyp24a1 (Fig. 3-4B), although renal Cyp24a1 induction was more sustained (maintained above 40-fold of basal level) than that of the ileum, whose Cyp24a1 levels rapidly returned to basal levels at 24 h (Fig. 3-3B). There are perhaps two potential explanations. First, enterocytes in the intestine have a much higher turnover rate than renal tubular cells (Ferraris et al., 1992; Bonventre, 2003), reducing the sustainability of ileal Cyp24a1 induction. Second, renal and not ileal VDR was induced, and renal

induction of VDR continued upon multiple dosing of 1,25(OH)2D3 (Fig. 3-4B), resulting in higher expression of renal Cyp24a1 mRNA. Visually, a correlation could be identified between renal VDR and Cyp24a1 mRNA levels (Fig. 3-4B) although no correlation was noted between ileal Cyp24a1 mRNA and VDR due to the small change in VDR (Fig. 3-3B).

The induction pattern of renal Mdr1 mRNA (Fig. 3-5B), another VDR target gene (Saeki et al., 2008), was similar to that of renal VDR mRNA (Fig. 3-4A). Induction of both Mdr1 mRNA and

P-gp protein was sustained in the kidney after consecutive injections of 1,25(OH)2D3 (Figs. 3-5, B and C). A higher P-gp protein expression (2.7-fold increase) was observed previously for the enhanced renal but not intestinal excretion of digoxin, a P-gp substrate, when the same

1,25(OH)2D3 doses were administered to the mouse (Chow et al., 2011a), as found in the present study (Fig. 3-5C). Thus, VDR regulates Mdr1 and P-gp induction in a tissue-specific manner.

Plasma calcium levels (Fig. 3-8D) were greatly influenced by temporal changes in mRNA and

protein expression of Trpv6, which was increased by 1,25(OH)2D3 treatment in both the intestine and kidney (Figs. 3-3C, 3-4C, and 3-8C). The calcium channel Trpv6 together with Trpv5 mediate the transcellular calcium transport following binding to calbindin to facilitate calcium diffusion across the basolateral membrane and extrusion via the ATP-dependent Ca2+-ATPase, PMCA1b, and Na+/Ca2+ exchanger NCX1 (Hoenderop et al., 2005). Trpv6 is a major contributor for the apical, intestinal absorption of calcium, since a lack of Trpv6 in knockout mice resulted in significant reduction in calcium absorption and plasma calcium levels (Bianco et al., 2007; Cui et al., 2012). Calcium balance is intimately related to Trpv5/6, whose channel activities are stimulated by the VDR primarily via the genomic transcription of the VDREs and by the estrogen receptor, ERα (Hoenderop et al., 2005). The increase in Trpv6 mRNA and protein expression in the mouse intestine and kidney strongly correlates with the increase (10-40%) in plasma calcium (Figs. 3-3C, 3-4C, 3-8C, and 3-8D) that in turn attenuated plasma PTH level (Fig. 3-9B) under our

dosing regimen of 1,25(OH)2D3. Elevated plasma calcium is expected to activate the CaSR in the parathyroid gland to inhibit synthesis of PTH, which in turn reduces the mRNA expression of renal Cyp27b1 (Jones et al., 1998; Feldman et al., 2005). These immediate changes were found in

response to the single and multiple 1,25(OH)2D3 doses in our study (Fig. 3-4D). Changes in ileal and renal Trpv6 protein expression and calcium (Figs. 3-8, C and D) indirectly regulated renal Cyp27b1 (Fig. 3-4D) via reduction of PTH (Fig. 3-9). Moreover, the higher Trpv6 protein content

in the duodenum when compared with that in kidney (Fig. 3-8B) suggests that the intestine is the more important organ than the kidney with respect to Trpv6 induction and calcium absorption (Fig. 3-8D).

Treatment of 1,25(OH)2D3 is known to result in both genomic and nongenomic effects (Revelli et al., 1998), as noted for Cyp24 (Hedlund et al., 1996; Nutchey et al., 2005). However, nongenomic effects are difficult to monitor, since these effects occur rapidly (from seconds to a few minutes), exemplified by the opening of calcium channels for calcium influx without changes in gene or

protein (Revelli et al., 1998). We believe that chronic treatment of 1,25(OH)2D3 will result in VDR effects that are mostly genomic, since there are notable changes in mRNA and protein levels of VDR target genes, namely, VDR, Cyp24a1, Trpv6, and Mdr1/P-gp (Figs. 3-3 to 3-8). Then we

examined the correlation between 1,25(OH)2D3 levels vs. mRNA expression of VDR target genes.

Our data revealed that the concentrations of 1,25(OH)2D3 in plasma and tissues peaked at 0.5 to 1

h (Fig. 3-2) and decayed rapidly with a t1/2 of ~6 h (Table 3-2), and induction of mRNA expression of VDR target genes in the ileum and kidney peaked at 3-9 h (Figs. 3-3 and 3-4), but there was no distinct pattern for protein expressions for many of the VDR target genes (Figs. 3-6, 3-7, and 3-8). In sum, the derived correlation was not always meaningful. A positive correlation was expected

between the 1,25(OH)2D3 and calcium levels in plasma if nongenomic effects had prevailed; however, a negative correlation was observed (data not shown). As expected, the correlation failed to divulge information on VDR mechanisms. The transduction processes are a multistep phenomenon that may involve multiple organs and multiple feedback or feed-forward loops, which is quite complex, especially with respect to calcium elevation.

Data from this study conclusively show that 1,25(OH)2D3 enters tissues rapidly, shown by the

parallel disposition profiles. The terminal half-life and the area under the curve (AUC0-48) of total

1,25(OH)2D3 (administered + endogenous) remained relatively unchanged between doses, and no trend was identifiable (Table 3-2 and Fig. 3-2). The lack of pharmacokinetic changes in drug exposure in these mouse studies is attributed to the immediate, maximal changes in the degradation and synthetic enzymes, Cyp24a1 and Cyp27b1, by VDR activation after the first administration of

1,25(OH)2D3. Hence, repeated dosing renders similar pharmacokinetic effects. A pattern may be

discerned for the temporal changes in tissue where concentrations of 1,25(OH)2D3 are correlated to temporal changes in the expression of some VDR target genes. VDR activation increased Trpv6

expression, which was higher in intestine and less so in kidney. Trpv6 is also responsible for the

sustained elevation of calcium and attenuation of PTH in plasma upon repeated 1,25(OH)2D3 dosing. With the higher, prevailing calcium concentration, decreased plasma PTH level and

Cyp27b1 expression in kidney ensue, rendering a lower synthesis of 1,25(OH)2D3. Rapid induction of Cyp24a1 mRNA and protein expression in kidney hastens 1,25(OH)2D3 clearance, evidenced

by plasma 1,25(OH)2D3 falling below basal levels at 24 h after single and chronic 1,25(OH)2D3

dosing. The temporal relationships between VDR target genes and 1,25(OH)2D3 levels in tissues and the dose- and route-dependency are currently under investigation with a mechanism-based pharmacokinetic/pharmacodynamic model.

3.6 Acknowledgments

We thank Matthew R. Durk for assistance in harvesting the tissues, and Dennis Wagner, Department of Nutritional Science, University of Toronto, for advice on the EIA assay. This work was supported by research grants from the Canadian Institutes for Health Research. Edwin C.Y. Chow was a recipient of the University of Toronto Open Fellowship and the National Sciences and the Engineering Research Council of Canada, Alexander Graham Bell Canada Graduate Scholarship (NSERC-CGS). Holly P. Quach was a recipient of the NSERC-CGS and Ontario Graduate Scholarship (OGS) fellowships.

3.7 Statement of significance of Chapter 3

In this chapter, concentrations of 1,25(OH)2D3 in plasma, kidney, intestine, and bone were measured and correlated to changes in VDR target genes in tissues. Not surprisingly, we found

that 1,25(OH)2D3 concentrations were much higher than basal levels upon administration of

1,25(OH)2D3. Interestingly, the PK parameters of 1,25(OH)2D3 remained unchanged over repeated dosing, a consequence of the induction of Cyp24a1 and inhibition of Cyp27b1, feedback

mechanisms that help maintain 1,25(OH)2D3 and calcium levels in homeostasis. This observation served as the basis behind the PKPD model that was constructed and presented in Chapter 7.

Here, we demonstrated that exogenously administered 1,25(OH)2D3 could rapidly equilibrate into target tissues to activate the VDR to exert direct effects. Changes in VDR and VDR target genes such as Cyp24a1, Trpv6, and Mdr1 correlated well to tissue 1,25(OH)2D3 concentrations. Hence, this study conclusively showed that the induction of VDR target genes is dependent on the treatment regimen and the resultant tissue concentrations of 1,25(OH)2D3.

My contributions to this work included administration of 1,25(OH)2D3 to mice, tissue harvesting,

mRNA and protein analyses, measurement of plasma and tissue 1,25(OH)2D3, plasma calcium analysis, and writing of the manuscript. Edwin C.Y. Chow, the first author, was instrumental in developing the EIA assay for our laboratory. He also contributed to all of these mentioned tasks and additionally measured the plasma PTH levels and performed the non-compartmental PK analysis. Reinhold Vieth contributed to the study design and gave recommendations in writing the paper. K. Sandy Pang contributed to the study design and writing of the paper.

Chapter 4

Vitamin D Receptor Down-Regulates the Small Heterodimer Partner and Increases CYP7A1 in a Mechanism Independent of the Farnesoid X Receptor

Contributions of H. P. Quach excerpted from:

Vitamin D Receptor Activation Down-Regulates the Small Heterodimer Partner and Increases CYP7A1 to Lower Cholesterol

Edwin C.Y. Chow1, Lilia Magomedova1, Holly P. Quach1, Rucha Patel1, Matthew R. Durk1, Jianghong Fan1, Han-Joo Maeng1, Kamdi Irondi1, Sayeepriyadarshini Anakk2, David D. Moore2, Carolyn L. Cummins1, and K. Sandy Pang1

1Department of Pharmaceutical Sciences, Leslie Dan Faculty of Pharmacy, University of Toronto, Toronto, Ontario, Canada

2Department of Molecular and Cellular Biology, Baylor College of Medicine, Houston, Texas, USA

Gastroenterology 2014; 146:1048-1059.

4.1 Abstract

Little is known about effects of the vitamin D receptor (VDR) on hepatic activity of cholesterol 7α-hydroxylase (CYP7A1) and cholesterol metabolism. We studied these processes in mice in vivo in two sets of studies. In the first study, C57BL/6 wild-type mice fed normal diets were given

intraperitoneal injections of 1α,25-dihydroxyvitamin D3 [1,25(OH)2D3; 4 doses, 2.5 μg/kg, every other day] and blood and livers were collected at different time points to assess whether

1,25(OH)2D3 could enter liver tissues to exert direct VDR effects. 1,25(OH)2D3 levels were measured by enzyme-immunoassay. In the second study, farnesoid X receptor knockout [Fxr(-/-)] mice were fed normal or Western diets for 3 weeks and then treated with 1,25(OH)2D3 (4 doses, 2.5 μg/kg, every other day during the last week of diet). Plasma and tissue samples were collected and levels of Vdr, small heterodimer partner (Shp), Cyp7a1, degradation enzyme (Cyp24a1), and fibroblast growth factor 15 (Fgf15) expression were measured using quantitative PCR and Western blotting. Plasma and liver cholesterol levels were measured using enzymatic methods. In mice fed normal diets and given injections of 1,25(OH)2D3, liver and plasma concentrations of 1,25(OH)2D3 increased and decreased in unison. Changes in hepatic Cyp7a1 mRNA correlated with those of Cyp24a1 (a Vdr target gene) and inversely to Shp mRNA, but not ileal Fgf15 mRNA. In Fxr(-/-) mice with hypercholesterolemia, injection of 1,25(OH)2D3 consistently reduced levels of plasma and liver cholesterol and Shp mRNA, and increased hepatic Cyp7a1 mRNA and protein; these

changes were not observed in Shp(-/-) mice given 1,25(OH)2D3 and fed Western diets. In mice, activation of the VDR directly represses hepatic SHP to increase levels of CYP7A1 and reduce cholesterol, a novel mechanism that was found to be independent of the FXR.

4.2 Introduction

Cholesterol is an essential component of cell membranes and the precursor to steroid hormones and bile acids. In excess, cholesterol can lead to atherosclerosis and coronary heart disease. In liver, cholesterol is metabolized to bile acids by cholesterol 7α-hydroxylase (CYP7A1), which is the rate-limiting metabolic enzyme in the classic bile acid synthetic pathway (Chiang, 2004). The CYP7A1 promoter contains highly conserved bile acid responsive regions known to be modulated by the feedback repression by various transcription factors in response to increasing hepatic bile 58

acid concentrations (Chiang, 2004). A primary, negative feedback mechanism of CYP7A1 regulation is the farnesoid X receptor (FXR, NR1H4) and small heterodimer partner (SHP, NR0B2) regulatory cascade (Lu et al., 2000). Bile acids such as chenodeoxycholic acid (CDCA) activate FXR to increase transcription of SHP, an atypical nuclear receptor that lacks a DNA binding domain and represses CYP7A1 activation by suppression of transcription factors, the liver receptor homolog-1 (LRH-1, NR5A2) and hepatocyte nuclear factor 4α (HNF-4α, NR2A1), which are essential for CYP7A1 expression (Zollner et al., 2006). A second negative feedback mechanism on CYP7A1 is found in the intestine, where activation of FXR induces the fibroblast growth factor 15/19 (rodent/human), a hormonal signaling molecule that represses CYP7A1 through interaction with the liver fibroblast growth factor receptor 4 (FGFR4) via the c-Jun signaling pathway (Inagaki et al., 2005).

The vitamin D receptor (VDR, NR1I1) binds to its endogenous ligand, 1α,25-dihydroxyvitamin

D3 [1,25(OH)2D3] or lithocholic acid (LCA, alternate VDR ligand) (Makishima et al., 2002) to activate the transcription of genes. Various mechanisms implicate a role for VDR on CYP7A1 regulation (Wagner et al., 2010). Activation of VDR was found to antagonize the CDCA- dependent transactivation of FXR in VDR-transfected HepG2 cells (Honjo et al., 2006) and blunt the Lxrα-mediated induction of Cyp7a1 mRNA in rat hepatoma cells (Jiang et al., 2006). VDR inhibition of CYP7A1 transcription in human hepatocytes and HepG2 cells has been attributed to blockage of HNF-4α-mediated activation of CYP7A1 (Han and Chiang, 2009). Induction of

intestinal Fgf15 following a high dose of 1,25(OH)2D3 in mice was shown to down-regulate Cyp7a1 mRNA level (Schmidt et al., 2010). By contrast, Vdr-knockout mice are reported to have

higher total serum cholesterol (Wang et al., 2009), and treatment with 1α-hydroxyvitamin D3, a potent 1,25(OH)2D3 precursor, up-regulated Cyp7a1 mRNA expression in mice (Nishida et al., 2009; Ogura et al., 2009). Likewise, doxercalciferol, a vitamin D analogue, decreased the accumulation of triglycerides and cholesterol in murine kidney (Wang et al., 2011). In rat,

1,25(OH)2D3 down-regulated liver Cyp7a1 by an indirect mechanism (Chow et al., 2009). The divergent views are partially reconciled by species differences, with VDR protein levels being extremely low in rat liver, but present at detectable levels in mouse and man (Gascon-Barre et al.,

2003). The downregulation of Cyp7a1 after 1,25(OH)2D3 treatment in the rat is explained as a secondary hepatic Fxr and not Vdr effect, since increased bile acid absorption into portal blood occurred as a result of Vdr-mediated induction of the intestinal apical sodium-dependent bile acid 59

transporter (Asbt) due to the absence of the Fxr-Shp-Lrh-1 pathway that normally ablates Asbt activity (Chow et al., 2009). Clinical reports relating vitamin D status to cholesterol [cholesterol

levels vs. 1,25(OH)2D3 or its less active precursor, 25-hydroxyvitamin D3, 25(OH)D3] are equivocal. Both short (Ponda et al., 2012) and long (Zittermann et al., 2009) term vitamin D treatment did not improve lipid profiles nor lower cholesterol and only resulted in slightly lower serum triglyceride, whereas other studies documented increased HDL-C (Salehpour et al., 2012), or decreased total cholesterol and triglyceride but not changed HDL-C and LDL-C levels (Rahimi- Ardabili et al., 2013). A recent, population-based study established an association between

1,25(OH)2D3 and HDL-C and 25(OH)D3 and total cholesterol, LDL-C, and triglyceride (Karhapaa et al., 2010). Atorvastatin, supplemented with vitamin D, further lowered total cholesterol and LDL-C (Schwartz, 2009), and patients receiving statin or niacin supplemented with vitamin D and fish oil showed reduced LDL-C and triglycerides and elevated HDL-C (Davis et al., 2009). Hence, the role of the VDR in liver cholesterol regulation remains controversial.

The intent of this study was to clarify the impact of Vdr on Cyp7a1 regulation and cholesterol

lowering. We demonstrated that 1,25(OH)2D3 given to mice rapidly reached the liver, resulting in induction of hepatic Cyp7a1 via downregulation of Shp. We showed that Vdr-mediated Shp repression and up-regulation of Cyp7a1 was Fxr-independent in vivo.

4.3 Materials and methods

4.3.1 Materials

1,25(OH)2D3 was purchased from Sigma-Aldrich Canada (Mississauga, ON). Antibodies against Cyp7a1 (N-17) were from Santa Cruz Biotechnology (Santa Cruz, CA), while Gapdh (6C5) was from Abcam (Cambridge, MA). Male C57BL/6 mice were obtained from Charles River (Senneville, QC) and Fxr(-/-) mice (C57BL/6 background) was from Dr. Frank J. Gonzalez (National Institutes of Health, Bethesda). Studies were performed in accordance with institutionally approved animal protocols.

4.3.2 1,25(OH)2D3 treatment of mice in vivo

In the first study, blood samples and livers from wild-type C57BL/6 mice were collected at

different time points during the 1,25(OH)2D3 treatment period, as previously described (Chow et al., 2013b). In the second study, Fxr(-/-) mice (8-12 weeks old; n = 4-10) were fed a normal or Western diet [Harlan Teklad Cat#88137; high fat (42%)/high cholesterol (0.2%) diet] for 3 weeks

and treated with 0 or 2.5 µg/kg 1,25(OH)2D3 i.p. every other day for 8 days (q2d x4; Day 0 to Day 8) during the last week of the diet. On Day 8, blood samples and tissues were obtained under anesthesia (Chow et al., 2009; Chow et al., 2011a).

4.3.3 Plasma analyses of calcium, phosphorus, ALT, and bile acids

Plasma was diluted 350-fold with 1% HNO3 and calcium and phosphorus were determined by inductively coupled plasma atomic emission spectroscopy (ICP-AES). Plasma alanine aminotransferase (ALT) and bile acid concentrations in the portal blood were determined by the ALT kit (Bioquant, Nashville, TN) and the total bile acids assay kit (Diazyme, Poway, CA), respectively, according to manufacturer’s protocol.

4.3.4 Determination of plasma and liver 1,25(OH)2D3

1,25(OH)2D3 concentrations in mouse plasma and liver were assayed using an enzyme- immunoassay kit (Chow et al., 2013b). Briefly, weighed liver samples were added to double- distilled water up to 1 ml. The sample was homogenized with 3.75 ml of a mixture of methylene chloride and methanol (1:2 vol/vol). Then 1.25 ml of methylene chloride was added, mixed for 1 min, followed by addition of 1.25 ml of double-distilled water and mixed for another minute, before centrifugation at 3000 rpm for 20 min at room temperature. The extractant (bottom phase) was retrieved by a glass syringe-metal needle set. The extraction procedure was repeated with another 1.25 ml of methylene chloride. The recovered, bottom extractant was pooled with that

from the previous extraction, dried under N2, and reconstituted in 0.3 ml of charcoal-stripped human serum and analyzed by the enzyme-immunoassay kit (Immunodiagnostics Systems Inc., Scottsdale, AZ).

4.3.5 Quantitative real-time PCR

Total mRNA extraction and qPCR procedures are previously described (Chow et al., 2009). mRNA expression was normalized to Cyclophilin for liver and kidney samples and to Villin for ileal samples, and then expressed as relative mRNA expression of the respective normal diet controls. Primer sequences are presented below (Table 4-1).

Table 4-1. Mouse primer sets for quantitative real-time PCR

Gene Bank Forward (5’→ 3’ Sequence) Reverse (5’→ 3’ Sequence) Number Vdr NM_009504 GAGGTGTCTGAAGCCTGGAG ACCTGCTTTCCTGGGTAGGT Fxr NM_009108 CGGAACAGAAACCTTGTTTCG TTGCCACATAAATATTCATTGAGATT Shp NM_011850 CAGCGCTGCCTGGAGTCT AGGATCGTGCCCTTCAGGTA Cyp7a1 NM_007824 AGCAACTAAACAACCTGCCAGTACTA GTCCGGATATTCAAGGATGCA Fgf15 NM_008003 ACGGGCTGATTCGCTACTC TGTAGCCTAAACAGTCCATTTCCT Asbt NM_011388 GATAGATGGCGACATGGACCTC CAATCGTTCCCGAGTCAACC Lrh-1 NM_001159769 CCCTGCTGGACTACACGGTTT CGGGTAGCCGAAGAAGTAGCT Lxrα NM_013839 GGATAGGGTTGGAGTCAGCA GGAGCGCCTGTTACACTGTT Hnf-4α NM_008261 CCAAGAGGTCCATGGTGTTTAAG GTGCCGAGGGACGATGTAGT ApoE NM_009696.3 AAGCAACCAACCCTGGGAG TGCACCCAGCGCAGGTA Vldlr NM_001161420.1 GAGCCCCTGAAGGAATGCC CCTATAACTAGGTCTTTGCAGATATGG Ldlr NM_010700.2 AGGCTGTGGGCTCCATAGG TGCGGTCCAGGGTCATCT Sr-b1 NM_016741.1 GGGAGCGTGGACCCTATGT CGTTGTCATTGAAGGTGATGT Abca1 NM_013454.3 CGTTTCCGGGAAGTGTCCTA CTAGAGATGACAAGGAGGATGGA Abcg5 NM_031884.1 TCAATGAGTTTTACGGCCTGAA GCACATCGGGTGATTTAGCA Abcg8 NM_026180.2 TGCCCACCTTCCACATGTC ATGAAGCCGGCAGTAAGGTAGA Bsep NM_021022.3 ACAGCACTACAGCTCATTCAGAG TCCATGCTCAAAGCCAATGATCA Ntcp NM_011387.2 ATCTGACCAGCATTGAGGCTC CCGTCGTAGATTCCTTGCTGT Cyp8b1 NM_010012.3 GCCTTCAAGTATGATCGGTTCCT GATCTTCTTGCCCGACTTGTAGA Cyp27a1 NM_024264.4 CTGCGTCAGGCTTTGAAACA TCGTTTAAGGCATCCGTGTAGA HMG CoA NM_008255.2 CAAGGAGCATGCAAAGACAA GCCATCACAGTGCCACATAC Reductase Ibabp NM_008375.1 CAAGGCTACCGTGAAGATGGA CCCACGACCTCCGAAGTCT Npc1l1 NM_207242.2 TGGACTGGAAGGACCATTTCC GCGCCCCGTAGTCAGCTAT Ostα NM_145932.3 TACAAGAACACCCTTTGCCC CGAGGAATCCAGAGACCAAA Ostβ NM_178933.2 GTATTTTCGTGCAGAAGATGCG TTTCTGTTTGCCAGGATGCTC Cyp24a1 NM_009996 CTGCCCCATTGACAAAAGGC CTCACCGTCGGTCATCAGC Trpv6 NM_022413 ATCGATGGCCCTGCCAACT CAGAGTAGAGGCCATCTTGTTGCTG Mrp4 NM_001163676.1 GGTTGGAATTGTGGGCAGAA TCGTCCGTGTGCTCATTGAA Mdr1 NM_011076 TACGACCCCATGGCTGGATC GGTAGCGAGTCGATGAACTG Villin NM_009509 TCCTGGCTATCCACAAGACC CTCTCGTTGCCTTGAACCTC Cyclophilin X58990 GGAGATGGCACAGGAGGAA GCCCGTAGTGCTTCAGCTT

4.3.6 Western blotting

Cyp7a1 protein expression was determined by Western blotting methods, as described (Chow et al., 2009; Chow et al., 2011a), after loading of 50 µg total protein samples onto 10% SDS- polyacrylamide gels and transferring to nitrocellulose membranes.

4.3.7 Plasma and liver cholesterol

Total plasma cholesterol was determined by the Total Cholesterol Kit (Wako Diagnostics, Richmond, VA). For liver cholesterol measurements, lipids were extracted from ~0.2 g liver, homogenized in chloroform:methanol (2:1, vol/vol), as previously described (Patel et al., 2011), and cholesterol concentrations were determined from extracts using Infinity Cholesterol reagents (Thermo Scientific, Rockford, IL).

4.3.8 Statistics

Data are expressed as mean ± SEM. For comparison between two groups, the Mann-Whitney U- test was used and P < 0.05 was set as the level of significance.

4.4 Results

4.4.1 Plasma calcium is increased upon 1,25(OH)2D3 treatment

Fxr(-/-) mice fed a Western diet had a ~25% increase in body weight compared with normal diet controls, while mice treated with 1,25(OH)2D3 had ~25% decrease in body weight compared with Western diet controls. Mice fed a Western diet had ~40% increase in plasma calcium compared with normal diet controls, while mice displayed ~30% increase in calcium when treated with

1,25(OH)2D3 vs. Western diet controls. There was no dramatic change in plasma phosphorous and ALT levels, and serum bile acid concentrations in the portal blood were non-significantly

increased with 1,25(OH)2D3 treatment (Table 4-2).

Table 4-2. Body weight and plasma calcium, phosphorus, ALT, and bile acid concentrationsa

Normal Diet Western Diet

Fxr(-/-) Vehicle-Treated Vehicle-Treated 1,25(OH)2D3- Control Control Treated Body weightb (g) 27.0 ± 0.9 33.7 ± 1.0† 24.8 ± 1.1* Plasma calcium (mg/dl) 8.1 ± 0.4 11.4 ± 0.4† 17.1 ± 0.6* Plasma phosphorus (mg/dl) 19.8 ± 1.3 20.9 ± 0.8 21.6 ± 0.9 Plasma ALT (IU/ml) 96.3 ± 9.0 81.4 ± 59.9 89.2 ± 54.3 Portal bile acids (μM) 27.6 ± 6.5 62.7 ± 10.0† 80.1 ± 25.5 a Data are represented as mean ± SEM (n = 4-8). b Body weight of mice before sacrifice. † P < 0.05 compared to Normal diet vehicle-treated control (Mann-Whitney U-test). * P < 0.05 compared to Western diet vehicle-treated control (Mann-Whitney U-test).

4.4.2 Parallel changes in liver 1,25(OH)2D3 concentrations and temporal Cyp7a1 mRNA and protein expression in wild-type mice

To further examine Cyp7a1 expression changes in response to Vdr activation, we systematically

analyzed the relationship between liver 1,25(OH)2D3 concentration and gene expression in normal

diet-fed wild-type mice throughout the 1,25(OH)2D3 treatment period. A diurnal variation was found in both basal Cyp7a1 mRNA and protein levels (insets of Figs. 4-1, B and C), with peaks occurring at around 9 p.m. and 12 a.m., respectively. After 1,25(OH)2D3 dosing, a biphasic decay

profile of liver 1,25(OH)2D3 that closely paralleled the plasma concentration-time curve (Chow et al., 2013) was observed (Fig. 4-1A). Cyp7a1 mRNA expression rose maximally at around 12 h post-injection, and levels were amplified with subsequent injections (60 and 80-fold higher; Fig. 4-1B). Patterns of Cyp7a1 mRNA and protein induction were similar (Figs. 4-1, B and C). The

increase in Cyp7a1 mRNA expression in response to 1,25(OH)2D3 was much higher (60-80-fold) than the peak of the circadian rhythm (6.7-fold, Fig. 4-1B).

Figure 4-1. Correlation between liver 1,25(OH)2D3 concentration and hepatic and Cyp7a1 mRNA and protein expression in normal diet-fed wildtype mice. Repeated administration of 1,25(OH)2D3 resulted in (A) biphasic decay of 1,25(OH)2D3 concentration in liver (mean points; n = 2-4; solid and open circles for levels in treated livers or basal levels in vehicle-treated livers) that paralleled those in plasma [grey solid and open symbols for treated and basal levels in vehicle- treated livers; previously published (Chow et al., 2013)] and corresponding changes in hepatic (B) Cyp7a1 mRNA and (C) Cyp7a1 protein expression (n = 2-4). The insets show the diurnal variation of basal Cyp7a1 mRNA and protein expression, peaking at around 9 p.m. and 12 a.m., respectively, in vehicle-treated mice. For (B) and (C), each point represents datum from one mouse, except for insets where the open symbols denote mean values (n = 2-4).

4.4.3 In hypercholesterolemic Fxr(-/-) mice, 1,25(OH)2D3 reduced plasma and liver cholesterol

To examine whether Fxr was involved in cholesterol lowering, we treated Fxr(-/-) mice fed a

normal or Western diet with 1,25(OH)2D3. Western diet alone did not alter basal Cyp7a1 levels (Fig. 4-2A) but increased plasma and liver cholesterol concentrations in Fxr(-/-) mice (Fig. 4-2B). When compared to normal diet-fed controls, Western diet-fed controls displayed higher expression of basal hepatic Shp mRNA and ileal Fgf15 mRNA, while expression of ileal Asbt mRNA was slightly lower (Fig. 4-2C).

When compared to Western diet-fed controls, 1,25(OH)2D3 treatment increased Cyp7a1 mRNA and protein expression (Fig. 4-2A). Accordingly, plasma and liver cholesterol were reduced with

1,25(OH)2D3 treatment (Fig. 4-2B). Upon treatment with 1,25(OH)2D3, hepatic Shp and ileal Fgf15 mRNA expression were reduced, while Asbt mRNA expression returned to basal levels in (Fig. 4-2C).

Figure 4-2. 1,25(OH)2D3 treatment increases hepatic Cyp7a1 and decreases hepatic Shp mRNA expression and plasma and liver cholesterol in Western diet-fed Fxr(-/-) mice. 1,25(OH)2D3-treatment (A) increased Cyp7a1 mRNA and protein expression, (B) decreased plasma and liver cholesterol concentrations, and (C) attenuated hepatic Shp, ileal Fgf15, and increased ileal Asbt mRNA levels in Western diet-fed mice (n = 4-8); P < 0.05: †Western diet vs. normal diet; * Western diet-fed, vehicle-treated control vs. Western diet-fed, 1,25(OH)2D3-treated mice (Mann-Whitney U-test).

4.4.4 mRNA expression in liver, ileum, and kidney

In the liver, mRNA expression levels of Vdr, Lxrα, Lrh-1, apolipoprotein E (ApoE), bile salt export pump (Bsep), and cholesterol efflux transporters (Abca1, Abcg5, and Abcg8) were all increased by the Western diet whereas mRNA levels Cyp8b1, an alternate cholesterol metabolizing enzyme, and HMG CoA reductase, a cholesterol synthesis enzyme, were decreased (Fig. 4-3A). Upon

1,25(OH)2D3 treatment, mRNA expression levels of Lxrα, Hnf-4α, ApoE, Abca1, and Abcg8 were decreased in Western diet-fed mice (Fig. 4-3A). In the ileum, mRNA expression of Abca1, Abcg5, and Abcg8 was elevated by the Western diet, but returned to basal levels upon treatment of

1,25(OH)2D3, whereas mRNA expression of Niemann-Pick C1-Like 1 (Npc1l1), a mediator of cholesterol absorption, was slightly decreased by the Western diet (Fig. 4-3B). In the kidney, mRNA expression levels of Vdr and VDR target genes (Cyp24a1, Trpv6, Mrp4, and Mdr1) were all increased with 1,25(OH)2D3 treatment, as expected (Fig. 4-3C).

Figure 4-3. Changes in mRNA expression in the (A) liver, (B) ileum, and (C) kidney of normal and Western diet-fed Fxr(-/-) mice treated with 1,25(OH)2D3. Data represent the mean ± SEM (n = 4-8). P < 0.05: †Western diet vs. normal diet controls; *Western diet-fed, vehicle- treated control vs. Western diet-fed, 1,25(OH)2D3-treated mice (Mann-Whitney U-test).

4.5 Discussion

We observed cholesterol lowering in response to 1,25(OH)2D3 treatment, an effect associated with elevated Cyp7a1 mRNA and protein expression in hypercholesterolemic Fxr(-/-) mice. We showed that increased Cyp7a1 expression and activity was achieved via Vdr-repression of Shp

following steady-state treatment of 1,25(OH)2D3. Through extensive gene profiling of the ileum and liver of hypercholesterolemic mouse models, we investigated the involvement of other transporters, enzymes or nuclear receptors known to modulate cholesterol or bile acid processing.

Here, we found that 1,25(OH)2D3 treatment could decrease the expression of hepatic and ileal cholesterol efflux transporters (Abca1, Abcg5, and Abcg8), which may further contribute to the

1,25(OH)2D3-mediated regulation of cholesterol homeostasis (Figs. 4-3, A and B). We also

confirmed in the kidney, a VDR abundant tissue, that 1,25(OH)2D3 treatment was effective since there was significant induction of renal VDR target genes (Fig. 4-3C). All data point to the Fxr- independent and Shp-dependent mechanism whereby Vdr down-regulates Shp to increase Cyp7a1 in cholesterol lowering.

These findings contrast other reports on the downregulation of mouse hepatic Cyp7a1 mRNA due

to Fgf15 induction after a high dose of 1,25(OH)2D3 (Schmidt et al., 2010). Such divergent results could be explained by differences in the dose administered. The notion that VDR is inhibitory to CYP7A1 in human hepatocytes was based on absence of substantiating evidence on CYP7A1 protein/activity or cholesterol measurements, or timed-matched control samples (Han and Chiang, 2009); the observations were explained as genomic effects arising from the interaction between VDR and HNF-4α on the CYP7A1 promoter, or as non-genomic effects via activation of the ERK pathway (Han and Chiang, 2009; Han et al., 2010), though the latter was not reproduced in a different cell line (Wu et al., 2007). It may be argued that conclusions based solely on in vitro data are debatable since time-dependent gene stability exists, and acute cell-based studies are expected to be sensitive to differences in treatment times and conditions. Indeed, in our in vitro human hepatocyte studies, we showed time-dependent changes in CYP7A1 expression (Chow et al., 2014). The involvement of Shp/SHP may have been easily missed since SHP mRNA has a short half-life (<30 min) due to rapid proteasomal degradation that is under control of the ERK pathway

(Miao et al., 2009). We suggest that longer term effects of steady-state doses of 1,25(OH)2D3 observed in our in vivo studies likely represent physiologic responses. 68

In our molecular studies, we confirmed an interaction between the Shp promoter and Vdr protein in the ChIP assay. The addition of VDR ligand resulted in reduced activity of Shp/SHP promoters (Chow et al., 2014), explaining the observed cholesterol lowering in vivo. Chromatin remodelling

at the SHP promoter in response to 1,25(OH)2D3 is consistent with ligand-mediated repression, though the relevant co-repressor proteins involved in this process have yet to be identified. While our data are consistent with a direct role for VDR in hepatocytes, we cannot rule out that intercellular communication between resident liver cells is also occurring in vivo since VDR is also highly expressed and functional in stellate cells (Ding et al., 2013). We acknowledge that

additional 1,25(OH)2D3-liganded mechanisms may contribute to lowering plasma and liver cholesterol in vivo. Indeed, 1,25(OH)2D3 treatment of HL-60 macrophages may also reduce cholesterol by inhibiting HMG-CoA reductase activity and increasing ACAT activity leading to cholesteryl ester accumulation (Jouni et al., 1995).

The novel mechanism of up-regulation of CYP7A1 after 1,25(OH)2D3 treatment suggests that the VDR is a new therapeutic target for cholesterol lowering. However, the potential utility of this mechanism to treat hypercholesterolemia is limited due to the dose-limiting hypercalcemia of

1,25(OH)2D3 (Chow et al., 2013b) or its precursor, 1α-hydroxyvitamin D3 (Ogura et al., 2009). Use of dietary vitamin D for cholesterol lowering in humans remains somewhat uncertain since only

very low levels of 1,25(OH)2D3 are synthesized after ingestion. It is highly probable that vitamin D deficiency would affect cholesterol status (Wang et al., 2009). The interplay between the VDR and cholesterol homeostasis in humans requires continued investigation with non-hypercalcemic VDR ligands.

4.6 Acknowledgments

The authors (KSP, CC, ECYC, LM, HPQ, and MRD) gratefully acknowledge support from the Canadian Institutes of Health Research, the Natural Sciences and Engineering Research Council of Canada, and the Ontario Graduate Scholarship Program. SA and DDM were supported by CPRIT grant RP120138, and R. P. Doherty, Jr. – Welch Chair in Science Q-0022. The authors thank Drs Nan Wu and Martin Wagner, Baylor College of Medicine, Texas Medical Center, Houston, for assistance with studies on the bile acid pool sizes of Shp(-/-) mice.

4.7 Statement of significance of Chapter 4

Here, we have conclusively shown that exogenously administered 1,25(OH)2D3 can rapidly distribute to the liver where it can modulate down-stream targets of the VDR. Previously, it was identified that the FXR negatively regulates CYP7A1 via the induction of SHP, which acts by attenuating the stimulatory effects of LRH-1 and HNF-4α in the liver (Goodwin et al., 2000; Lu et al., 2000; Zollner et al., 2006), and by intestinal FXR through induction of FGF15 to inhibit CYP7A1 (Inagaki et al., 2005). In this chapter, we demonstrated in hypercholesterolemic Fxr(-/-)

mice that 1,25(OH)2D3-liganded VDR can inhibit hepatic Shp to remove its inhibition on Cyp7a1, thereby decreasing plasma and liver cholesterol levels. Meanwhile, these cholesterol lowering trends were not observed in Shp(-/-) mice (Chow et al., 2014). Therefore, the VDR acts in an FXR-

independent but SHP-dependent mechanism to regulate cholesterol levels. 1,25(OH)2D3-treated hypercholesterolemic Fxr(-/-) mice also displayed decreased mRNA expression of hepatic and ileal cholesterol efflux transporters (Abca1, Abcg5, and Abcg8), and thus, it cannot be ruled out

that 1,25(OH)2D3 and the VDR could act through multiple mechanisms to regulate cholesterol homeostasis in the body.

Since we have identified the VDR as a potential therapeutic target for cholesterol lowering, we further performed studies where we treated hypercholesterolemic wild-type mice with

1,25(OH)2D3 alone, atorvastatin (inhibitor of cholesterol synthesis) alone, or a combination of the two (Appendix I). In collaboration with Dr. Geny M.M. Groothuis (University of Groningen, The Netherlands), we also performed studies in human precision-cut liver slices exposed to

1,25(OH)2D3 to determine whether human CYP7A1 expression increased in a similar manner to that of mice (Appendix I).

My contributions to this chapter excerpted from Appendix V (Chow et al., 2014) include all of the

analyses from the Fxr(-/-) mice study, including 1,25(OH)2D3 treatment, tissue harvesting, mRNA and protein analyses, measurements of plasma calcium, phosphorus, and ALT, and bile acid concentrations in the portal blood, and writing of the manuscript. Along with Edwin C.Y. Chow,

I also contributed to the temporal study where we measured plasma and liver 1,25(OH)2D3 levels and Cyp7a1 mRNA and protein. Edwin C.Y. Chow, Lilia Magomedova, Rucha Patel, Matthew R. Durk, Carolyn L. Cummins, and K. Sandy Pang also contributed to writing of the manuscript.

Chapter 5

Vitamin D Deficiency Triggers Hypercholesterolemia That Is

Reversed Upon Treatment with 1α,25-Dihydroxyvitamin D3 or

Vitamin D3 in Mice

Holly P. Quach1, Stacie Y. Hoi1, Jie Chen1, Adrie Bruinsma2, Geny M.M. Groothuis2, Albert P. Li3, Edwin C.Y. Chow1, and K. Sandy Pang1

1Department of Pharmaceutical Sciences, Leslie Dan Faculty of Pharmacy, University of Toronto, Toronto, Ontario, Canada

2Division of Pharmacokinetics, Toxicology and Targeting, Department of Pharmacy, University of Groningen, Groningen, The Netherlands

3In Vitro ADMET Laboratories, Columbia, Maryland, USA

5.1 Abstract

Background: The relationship between vitamin D-deficiency and hypercholesterolemia is largely

unknown, although 1α,25-dihydroxyvitamin D3 [1,25(OH)2D3]-activated vitamin D receptor (Vdr) was found to repress the small heterodimer partner (Shp) and increase cholesterol 7α-hydroxylase (Cyp7a1/CYP7A1) expression and cholesterol metabolism in mice. Objectives: The impact of vitamin D-deficiency on cholesterol levels and altered Shp and Cyp7a1 expression was evaluated. Methods: A vitamin D-deficiency model was established in male C57BL/6 mice (3-week-old) that were fed a diet containing sufficient (VD+; 2200 IU/kg) or deficient (VD-; 0 IU/kg)

cholecalciferol (D3) for 8-weeks. These VD- diets contained 0.47% or 2.5% calcium. For intervention studies, mice fed VD+ or VD- diets for the entire 8- or 11-week duration were further placed on high fat/high cholesterol (HF/HC; 42% fat/0.2% cholesterol) VD+ or VD- diets during the last 3-weeks. VD- mice were treated intraperitoneally every other day during the last 1- or 4-

weeks with 1,25(OH)2D3 (2.5 µg/kg) or D3 (20 µg/kg), respectively. Plasma and liver cholesterol were measured enzymatically, bile acid pool size was determined by LC-MS/MS, and gene expression by qPCR and Western blotting. Human liver tissues were also examined for

correlations between 1,25(OH)2D3, cholesterol, and CYP7A1. Results: Compared with VD+, mice fed VD- diets (0.47% or 2.5% calcium) for 8-weeks exhibited decreased liver 1,25(OH)2D3 (90-93%; P<0.001), Vdr (41-59%; P<0.05) and Cyp7a1 (37-63%; P<0.05), but elevated Shp (0.5- 2.9-fold; P<0.0001) and liver cholesterol (22-30%; P<0.05). Negative correlations were found between liver cholesterol and 1,25(OH)2D3 (r=-0.77; P<0.001) or Cyp7a1 (r=-0.82; P<0.0001).

Similar correlations between cholesterol and 1,25(OH)2D3 (r=-0.56; P=0.07) or CYP7A1 (r=-

0.60; P=0.05) were found for human liver tissues. Treatment with 1,25(OH)2D3 and D3 reduced plasma (32-44%) and liver (27-59%) cholesterol in HF/HC VD- mice, with concomitant suppression of Shp (42-72%) and increased Cyp7a1 (0.96-3.9-fold) and bile acid pool sizes (26- 34%; P>0.05). Conclusions: Vitamin D-deficiency is correlated with hypercholesterolemia in mice.

5.2 Introduction

In the clinical setting, vitamin D status is determined by circulating levels of 25-hydroxyvitamin

D3 [25(OH)D3], and not its active metabolite, 1α,25-dihydroxyvitamin D3 [1,25(OH)2D3], the natural ligand of the vitamin D receptor (VDR) that exists at extremely low levels (Hollis, 1996). Vitamin D-deficiency is the consequence of inadequate sunlight exposure and/or insufficient dietary intake of vitamin D, and is further influenced by the latitude, season, age, skin pigmentation (Mithal et al., 2009), or genetic variations of the synthetic and degradation enzymes (25- hydroxylase/CYP27A1 and CYP2R1, 1α-hydroxylase/CYP27B1, and 24-hydroxylase/CYP24A1) for 25(OH)D3 and 1,25(OH)2D3 (Berry and Hypponen, 2011). The increasing prevalence of vitamin D-deficiency poses as a major health concern since there is growing evidence to suggest that this nutritional deficit contributes to the pathogenesis of cancer, metabolic syndrome, and cardiovascular diseases (Gorham et al., 2005; Dobnig et al., 2008; Ghanei et al., 2015).

Insufficient sunlight exposure is found to be associated with elevated serum cholesterol levels (Grimes et al., 1996), a major accelerator of coronary heart disease (Martin et al., 1986). Although there were a number of cross-sectional studies that showed strong correlations between vitamin D-deficiency and hypercholesterolemia (Karhapaa et al., 2010; Schnatz et al., 2014), there is little direct evidence to link the two. A potential role of vitamin D in hypercholesterolemia was inferred when total and low density lipoprotein (LDL) cholesterol levels in atorvastatin-treated patients

were further reduced upon supplementation with cholecalciferol (vitamin D3 or D3) (Schwartz,

2009). Associations were also observed between 25(OH)D3 deficiency and elevated total and LDL

cholesterol levels (Karhapaa et al., 2010), while supplementation with calcium and vitamin D3 increased both 25(OH)D3 and HDL cholesterol but reduced LDL cholesterol levels (Schnatz et al., 2014). Despite that these reports support the notion that high vitamin D concentration improves cholesterol profiles, contradictions have also been found (Heikkinen et al., 1997; Ponda et al., 2012). Hence, the relationship between vitamin D deficiency and hypercholesterolemia requires further exploration.

Regulatory events on cholesterol and bile acid metabolism have been well characterized (Makishima et al., 1999; Goodwin et al., 2000; Chiang, 2004). The synthesis of cholesterol, a precursor to steroid hormones and bile acids, is predominantly governed by the sterol regulatory 73

element-binding protein 2 (SREBP2)-mediated regulation of 3-hydroxy-3-methylglutaryl- coenzmye A reductase (HMGCR) in the liver, whereas metabolism is mediated by cholesterol 7α- hydroxylase (CYP7A1), the rate-limiting enzyme in the classic bile acid synthetic pathway (Chiang, 2004). A regulatory feedback mechanism of CYP7A1 is via the farnesoid X receptor (FXR)-small heterodimer partner (SHP) cascade in the liver (Goodwin et al., 2000). Bile acids, products of cholesterol metabolism via CYP7A1, are ligands of the FXR (Makishima et al., 1999) that increase the transcription of SHP, an atypical nuclear receptor that represses CYP7A1 activation by suppression of essential transcription factors (Goodwin et al., 2000; Lee et al., 2000). A second negative feedback mechanism exists with the intestinal hormone, fibroblast growth factor 15/19 (rodent Fgf15/human FGF19), which is activated by FXR and interacts with liver fibroblast growth factor receptor 4 to repress CYP7A1 (Inagaki et al., 2005).

The VDR has surfaced to be another key regulator of cholesterol and bile acid homeostasis, and elevated cholesterol levels were displayed in Vdr-knockout mice (Wang et al., 2009). The VDR was shown to increase Fgf15 in the suppression of Cyp7a1 in mice in vivo at high doses of

1,25(OH)2D3 (Schmidt et al., 2010) and antagonize FXR in vitro (Honjo et al., 2006). Smaller

doses of 1,25(OH)2D3 given repeatedly to mice suppressed Shp directly and independent of Fxr to increase Cyp7a1 expression and cholesterol metabolism, a mechanism that was elucidated in wildtype, Fxr(-/-), and Shp(-/-) mice (Chow et al., 2014). Mouse and human Shp/SHP promoter

activity studies revealed that 1,25(OH)2D3-mediated suppression of SHP occurred through the direct binding of VDR to vitamin D response elements in the SHP promoter (Chow et al., 2014).

Furthermore, human hepatocytes, when incubated with 1,25(OH)2D3, showed increased CYP7A1 mRNA and protein expression (Chow et al., 2014), showing that the VDR could play a critical role in human cholesterol homeostasis. Here, we hypothesize that vitamin D-deficiency increases cholesterol levels via downregulation of the Vdr-mediated suppression of Shp to reduce Cyp7a1

expression, while treatment with 1,25(OH)2D3 or D3 reverses these changes and restores cholesterol back to normal levels.

5.3 Materials and methods

5.3.1 Materials

Vitamin D3 (cholecalciferol) and 1,25(OH)2D3 powder were obtained from Sigma-Aldrich (Mississauga, ON). Mouse Cyp7a1 (N-17) (Santa Cruz Biotechnology, Santa Cruz, CA), Vdr (9A7), Gapdh (6C5), and human CYP7A1 (C-terminus) antibodies were purchased from Abcam (Cambridge, MA); donkey anti-goat (Jackson Immunoresearch Laboratories, West Grove, PA), goat anti-rat (Santa Cruz Biotechnology), goat anti-mouse, and goat anti-rabbit (Bio-Rad, Mississauga, ON) antibodies were also obtained. The Immunodiagnostics Systems enzyme-

immunoassay kits for 25(OH)D3 and 1,25(OH)2D3 measurements were obtained from Inter Medico (Markham, ON). The Immutopics parathyroid hormone (PTH) 1-84 ELISA kit was procured from Joldon Diagnostics (Burlington, ON). Bile acids and internal standards were obtained from Sigma-Aldrich, Steraloids (Newport, RI), and CDN Isotopes (Pointe-Claire, QC). All other reagents were from Sigma-Aldrich or Fisher Scientific (Mississauga, ON).

5.3.2 Human liver tissues

Human livers (n = 11) were provided by In Vitro ADMET Laboratories from the International Institute for the Advancement of Medicine (Edison, NJ) and the National Disease Research Interchange (Philadelphia, PA) (Supplementary Table S5-1).

5.3.3 Animal studies

5.3.3.1 Diets

Changes in cholesterol, 1,25(OH)2D3, and Vdr-related genes were examined with 3 diets to

establish vitamin D-deficiency. The vitamin D-sufficient (VD+; 2200 IU/kg D3, 0.47% Ca, 0.3%

P) diet was used as the control group. We also used two vitamin D-deficient (VD–; 0 IU/kg D3) diets, one without calcium (Ca; 0.47%) and phosphorus (P; 0.3%), and one re-supplemented with Ca (2.5%) and P (1.5%). The rationale for testing both VD- diets (without or with Ca) was due to the observation that low Ca levels could elevate Cyp27b1 expression and deplete 25(OH)D3 (Vieth et al., 1987) to promote 1,25(OH)2D3 synthesis (Goff et al., 1992; Song and Fleet, 2007). These 75

diets are all defined as the “normal diet”, and contrasted the 3-week high fat/high cholesterol (HF/HC; 42% fat/0.2% cholesterol) diet, which was shown to elevate plasma and liver cholesterol without changing Cyp7a1 mRNA or protein expression (Chow et al., 2014). The normal VD+ (TD.07370) and normal VD- diets, without (TD.89123) or with (TD.07541) Ca and P supplementation, and the HF/HC VD+ (TD.130146) and HF/HC VD- diets, without (TD.140244) or with (TD.130148) Ca and P supplementation, which were prepared by Harlan Laboratories (Madison, WI), were used in our studies. The composition is summarized in Supplementary Table S5-2.

5.3.3.2 Animal housing and tissue collection

All protocols were approved by the Animal Care Committee at the University of Toronto. Male C57BL/6 mice (Charles River Canada; Saint-Constant, QC) were maintained in groups of 3 or 4 mice per cage and given food and water ad libitum under a 12 h light/dark cycle; the VD- mice

were housed under incandescent light to minimize the endogenous production of vitamin D3. Mice were anesthetized with 150 and 10 mg/kg ketamine and xylazine, respectively, by intraperitoneal (i.p.) injection prior to cardiac puncture for blood collection with a 1 mL syringe-23G 19 mm needle-set that was pre-rinsed with heparin (1000 IU/mL). All mice were euthanized at 1200 h to minimize effects of Cyp7a1 circadian rhythm (Noshiro et al., 1990). For tissue collection, the abdominal vena cava was perfused with ice-cold saline to remove residual blood, and the liver, kidney, and scraped enterocytes were harvested, as described (Chow et al., 2011a). Samples were stored at -80°C until analyses.

5.3.3.3 Study 1: Establishment of the vitamin D-deficient mouse model

Mice (3-weeks old) were fed the normal VD+ (n = 3-4 per group) or normal VD- diet without (n = 3-4 per group) or with (n = 6-7 per group) Ca supplementation for 0-, 2-, 4-, 6-, or 8-weeks (Fig. 5-1A). By the end of each diet, mice were 11-weeks old (28.5±0.4 g).

5.3.3.4 Study 2: Intervention with short-term 1,25(OH)2D3

Mice were fed either the VD+ or VD- diet for 8-weeks, consisting of 5-weeks of normal diet followed by the normal or HF/HC VD+ or VD- diet for the next 3-weeks, a duration within which Cyp7a1 was known to remain unchanged and would not confound data interpretation (Chow et al., 76

2014) (Fig. 5-1B). The observation was confirmed (data not shown). Beginning on week 7 of the diets, mice fed the normal VD+ or HF/HC VD+ diets were treated i.p. with vehicle control (corn oil) (n = 3-4 per group), and mice fed the normal VD- or HF/HC VD- diets were treated i.p. with

vehicle control (corn oil) (n = 4 per group) or 2.5 μg/kg 1,25(OH)2D3 (n = 5-6 per group) every other day for 8 days at 1000 h, and euthanized 50 h after the last dose at 1200 h. Mice were 11- weeks old (29.5±0.4 and 32.6±0.7 g for normal and HF/HC diet fed mice, respectively) at the end of the study.

5.3.3.5 Study 3: Intervention with long-term vitamin D3

Preliminary studies showed that treatment of 20 μg/kg D3 i.p. every other day for 8 days did not

fully restore parameters back to baseline levels (data not shown). Hence a longer D3 treatment regimen (4-weeks) was used. Mice were fed either a normal VD+ diet for 11- weeks, or a normal VD+ diet for the first 8-weeks then a HF/HC VD+ diet for the remaining 3-weeks. In parallel, mice were fed a normal VD- diet for 11-weeks, or a normal VD- diet for 8-weeks then a HF/HC VD- diet for the remaining 3-weeks. Starting on week 7 of these diets, mice on the normal VD+ or HF/HC VD+ diets were treated i.p. with vehicle control (corn oil) (n = 4 per group), while mice fed the normal VD- or HF/HC VD- diets were treated i.p. with vehicle control (corn oil) (n = 5 per

group) or 20 μg/kg D3 (n = 6 per group) every other day during the last 4-weeks of diet (Fig. 5- 1C). Mice were euthanized for tissue collection at 50 h after the last dose. At the end of the study, mice were 14-weeks old (35.1±1.2 and 35.5±0.9 g for normal and HF/HC diet fed mice, respectively).

5.3.3.6 Bile acid pool size studies

Mice that were examined for bile acid pool sizes were placed on the respective diets and treated identically to the regimens for the short-term and long-term intervention studies, with n = 4 per group. The amounts and types of bile acids and bile acid pool sizes were determined and normalized to body weight, a metric that did not differ significantly between any groups (Tables A2-4 and A2-7; Appendix II). At the end of the study, mice were feed-deprived for 4 h (from 0800 to 1200 h) prior to anesthesia and the intact gall bladder, liver, and intestine were removed altogether, as previously described (Chow et al., 2014).

Figure 5-1. Experimental designs for (A) study 1: establishing the vitamin D-deficiency model, (B) study 2: short-term intervention with 1,25(OH)2D3, and (C) study 3: long-term intervention with D3 in mice. (A) Mice were fed a normal VD+ diet (white) or VD- diets without (grey) or with (black) Ca supplementation for 0- to 8-weeks. (B) Mice were fed a normal (white) VD+ or VD- diet for 8-weeks or placed on a HF/HC (grey) VD+ or VD- diet during the last 3- weeks. Beginning on week 7, mice were treated (hashed bar) with vehicle control (corn oil) or 2.5 µg/kg 1,25(OH)2D3 intraperitoneally every other day for 4 doses. (C) Mice were fed a normal (white) VD+ or VD- diet for 11-weeks or placed on a HF/HC (grey) VD+ or VD- diet during the last 3-weeks. Beginning on week 7, mice were treated (hashed bar) with vehicle control (corn oil) or 20 µg/kg D3 intraperitoneally every other day for 16 doses. The number of mice (n) per group is shown. Vertical arrows above diet bars represent treatment; horizontal arrows below diet bars represent euthanasia of mice for tissue collection.

5.3.4 Analysis of 25(OH)D3, 1,25(OH)2D3, calcium, and PTH

The concentrations of 25(OH)D3 and 1,25(OH)2D3 were assayed by enzyme-immunoassay kits, as

per manufacturer’s protocols. The extraction procedure and quantification of liver 1,25(OH)2D3 were identical to those previously described (Chow et al., 2013b). Plasma calcium was quantified by inductively coupled plasma atomic emission spectroscopy (Optima 3000 DV, Perkin Elmer Canada, Woodbridge, ON) (Chow et al., 2011a). PTH concentrations were determined by ELISA (Chow et al., 2013b).

5.3.5 Quantitative real-time PCR

Total RNA extraction and cDNA synthesis procedures were performed as described (Chow et al.,

2011a). The critical threshold cycle (CT) values of target genes for mouse liver and kidney were normalized to that for Cyclophilin, whereas those for intestinal samples were normalized to that for Villin, then expressed as relative mRNA expression of the 0-week or normal VD+ diet controls. Human liver genes were normalized to GAPDH. Primer sequences were identical to those previously reported (Chow et al., 2014).

5.3.6 Western blotting

Western blotting analysis was performed to determine protein expression of Vdr and Cyp7a1, normalized to GAPDH, as previously described (Chow et al., 2009; Chow et al., 2011a).

5.3.7 Determination of plasma and liver total cholesterol levels

Total plasma cholesterol levels were determined by the Total Cholesterol Kit (Wako Diagnostics, Richmond, VA). For liver cholesterol measurements, lipids were extracted from ~0.2 g liver tissue homogenized in chloroform:methanol (2:1 vol/vol) (Folch et al., 1957) and cholesterol was determined from extracts using Infinity Cholesterol Reagents (Thermo Scientific, Rockford, IL).

5.3.8 Determination of bile acids by LC-MS/MS

The extraction procedure for bile acids, with internal standards [a mixture of 0.25 mg/mL of CA-

d4, DCA-d4, CDCA-d4 and LCA-d4], to determine bile acid pool sizes, was identical to that

previously described (Chow et al., 2014). Bile acid standards (0.1-500 µM) were extracted in an identical fashion. The extracted samples were analyzed by LC-MS/MS (AB Sciex API 4000 Triple Quad LC/MS; Applied Biosystems, Burlington, ON) with an ESI source in negative ion mode, and the MS parameters that are summarized in Supplementary Table S5-3. The LC-MS/MS method used was similar to that of Alnouti et al. (2008), with modifications. Samples (10 µl) were separated on a Kinetex 2.6 µm C18 100A 100x4.6 mm column (Phenomenex Inc., Torrance, CA) and a Security Guard pre-column at 600 µl/min flow rate. The mobile phase consisted of pre- filtered 10 mM ammonium acetate (A) and HPLC grade acetonitrile (B). A gradient was utilized over 20 min: 0-6 min, 35-35% solvent B; 6-14 min, 35-58% solvent B; 14-15 min, 58-95% solvent B; 15-16.5 min, 95-95% solvent B; 16.5-17 min, 95-35% solvent B; 17-20 min, 35-35% solvent B. Bile acids were quantified based on area under the curve of the peak, corrected by the appropriate internal standard, using calibration curves based on authentic standards.

5.3.9 Statistics

Values are expressed as means ± SEMs. For comparison of means of the VD+, VD- (0.47% Ca), or VD- (2.5% Ca) diets at the same time point in Study 1, and for comparison of means in respective diet groups (normal or HF/HC) for intervention studies (Study 2 and Study 3), the one- way ANOVA and a post hoc Tukey honest significant difference test was performed with GraphPad Prism (version 6.01, GraphPad Software Inc., La Jolla, CA). We assessed the ANOVA model for assumptions of homogeneity of variance using the Brown-Forsythe test but no outlier was found to significantly influence the model. The unpaired Student’s t-test was used for comparison of normal VD+ and HF/HC VD+ diet controls (Study 2 and Study 3). Correlations were analyzed by Pearson’s correlation coefficient. A P value of < 0.05 was considered statistically significant.

5.4 Results

5.4.1 Study 1: Vitamin D deficiency mouse model with reduced levels of

25(OH)D3 and 1,25(OH)2D3

Mice fed the VD- diets (without or with Ca supplementation) displayed progressively and

significantly lower plasma 25(OH)D3 and 1,25(OH)2D3 (Figs. 5-2, A and B) and liver 1,25(OH)2D3 (Fig. 5-2C) levels when compared with their corresponding VD+ controls, and maximal changes occurred at 8-weeks. Unlike previous reports showing decreased plasma 25(OH)D3 (Vieth et al.,

1987) and increased plasma 1,25(OH)2D3 levels (Goff et al., 1992; Song and Fleet, 2007) in rodents with low plasma Ca, the absence or presence of supplemented Ca in the diet did not appear

to affect levels of 25(OH)D3 nor 1,25(OH)2D3 after 8-weeks of the VD- diets (Figs. 5-2, A to C).

Compared to VD+ diet, mice fed VD- diets displayed lower plasma and liver 1,25(OH)2D3 levels

(Supplemental Fig. S5-1A). The liver-to-plasma tissue partitioning ratio (Kp or Cliver/Cplasma) of

1,25(OH)2D3 was also unchanged between the VD+ and VD- mice (Supplemental Fig. S5-1B),

inferring that 1,25(OH)2D3 passively and rapidly equilibrated between liver and plasma and that

the liver tissue partitioning was unaffected by the vitamin D status. In kidney, basal 1,25(OH)2D3 levels were unchanged between VD+ and VD- mice (Supplemental Fig. S5-1A), but the renal Kp values differed significantly due to the low plasma 1,25(OH)2D3 levels (P<0.05) (Supplemental

Fig. S5-1B). The lack of change in renal 1,25(OH)2D3 is likely the result of increased plasma PTH

(Supplemental Fig. S5-1C) and induction/inhibition of the 1,25(OH)2D3 synthetic (Cyp27b1)/degradative (Cyp24a1) enzymes in mice fed VD- diets (Supplemental Figs. S5-1, D and E).

Figure 5-2. Reduced 25(OH)D3 and 1,25(OH)2D3 and elevated cholesterol levels in mice fed normal VD+ or VD- diets for 0- to 8-weeks. Mice were fed VD+ (white) or VD- diets without (grey) or with (black) Ca supplementation for 0- to 8-weeks. (A) Plasma 25(OH)D3, and (B) plasma and (C) liver 1,25(OH)2D3 levels were significantly reduced at the end of 8-weeks of the VD- diet. (D) Plasma Ca levels remained unchanged regardless of diet, while (E) plasma and (F) liver cholesterol levels were elevated after 8-weeks of VD- diet. Values are means ± SEMs, n = 3- 7 per group; labeled means at a time without a common letter differ, P<0.05. 82

5.4.2 Vitamin D deficiency increased cholesterol due to reduced Vdr and elevated Shp expression, thereby reducing Cyp7a1 expression

While plasma Ca levels were similar in mice fed the VD- diet without or with Ca supplementation throughout the duration of diet (Fig. 5-2D), plasma and liver cholesterol levels were significantly increased when compared with corresponding VD+ controls at the end of the 8-week diet (Figs. 5-2, E and F). The observed hypercholesterolemia was a consequence of deficiency-related perturbations of the cholesterol synthesis or metabolic pathways. In the liver, the mRNA expression of Srebp2 and Hmgcr, cholesterol synthesis regulators, was unchanged and increased non-significantly (P=0.42), respectively, after 8-weeks of VD- diet (Figs. A2-3, A and D; Appendix II). Vitamin D-deficiency, however, appeared to have altered the expression of regulators of cholesterol metabolism. In the liver, Vdr mRNA (P<0.05) and Vdr protein (P=0.10) expression were decreased with VD- diets, whereas both hepatic Fxr and Shp expression was increased, and Cyp7a1 mRNA and protein expression were slightly suppressed (Fig. 5-3A). By contrast, in the ileum, Vdr mRNA expression was unchanged, whereas Fxr (P=0.34) and Fgf15 (P<0.05) mRNA expression were non-significantly and significantly increased at 8-weeks, respectively (Fig. 5-3B).

In order to elucidate whether changes are related to vitamin D-deficiency, the expression of each gene was plotted against its corresponding liver 1,25(OH)2D3 measurement for each mouse. Here, we found no correlation between liver 1,25(OH)2D3 and Srebp2 or Hmgcr (Figs. A2-3, B and E; Appendix II), suggesting that cholesterol synthesis was not directly related to vitamin D-

deficiency. In contrast, significant, positive correlations were observed between liver 1,25(OH)2D3 and Vdr mRNA (Fig. 5-4A), Cyp7a1 mRNA (Fig. 5-4B), and Cyp7a1 protein (Fig. 5-4C).

Meanwhile, inverse correlations between Shp mRNA and liver 1,25(OH)2D3 (Fig. 5-4D), Cyp7a1 mRNA and Shp mRNA (Fig. 5-4E), and Cyp7a1 protein and Shp mRNA (Fig. 5-4F) were noted, suggesting that the Shp-mediated suppression of Cyp7a1 may be directly influenced by vitamin D-deficient conditions. Taken together, these findings reveal a direct association between vitamin D-deficiency and decreased cholesterol metabolism in mice. Furthermore, when liver cholesterol was expressed against 1,25(OH)2D3 (Fig. 5-5A), Cyp7a1 mRNA (Fig. 5-5B), and Cyp7a1 protein (Fig. 5-5C), strong inverse correlations were observed.

Figure 5-3. Changes in cholesterol-regulating genes in the liver and ileum with mice fed normal VD+ or VD- diets for 0- to 8-weeks. (A) In livers, mice fed VD- diets without (grey) or with (black) Ca supplementation showed slightly decreased Vdr mRNA and protein, elevated Fxr and Shp mRNA expression, and lower Cyp7a1 mRNA and protein expression after 8-weeks of diet, compared with the control, VD+ diet (white). (B) In ileum, the mRNA expression of Vdr remained unchanged, while Fxr and Fgf15 were non-significantly increased after 8-weeks of the VD- diets. Values are means ± SEMs, n = 3-7 per group; labeled means at a time without a common letter differ, P<0.05.

Figure 5-4. Changes in liver gene expression are correlated with liver 1,25(OH)2D3 levels in vehicle-treated mice fed normal VD+ or VD- diets. Vehicle-treated mice were fed VD+ (white) or VD- diets without (grey) or with (black) Ca supplementation for 8- or 11-weeks. At the end of the diets, liver 1,25(OH)2D3 levels were found positively correlated to (A) Vdr mRNA, (B) Cyp7a1 mRNA and (C) Cyp7a1 protein expression, while (D) Shp mRNA expression was inversely correlated. An inverse correlation also exists between (E) Shp mRNA and Cyp7a1 mRNA or (F) Shp mRNA and Cyp7a1 protein expression. Each data point represents one mouse.

Figure 5-5. Cholesterol is correlated with 1,25(OH)2D3 and Cyp7a1 expression in the mouse liver tissues of vehicle-treated mice fed normal VD+ or VD- diets. Vehicle-treated mice were fed VD+ (white) or VD- diets without (grey) or with (black) Ca supplementation for 8- or 11-weeks. At the end of diet, liver cholesterol was inversely correlated with (A) liver 1,25(OH)2D3 levels, (B) Cyp7a1 mRNA, and (C) Cyp7a1 protein expression in mice. Each data point represents one mouse. 85

5.4.3 Human liver cholesterol levels are inversely related to 1,25(OH)2D3 and CYP7A1 expression

In consonance with what was found in mouse liver tissue (Fig. 5-5), examination of human liver tissues revealed similar, although non-significant (P>0.05), inverse correlations between

cholesterol and 1,25(OH)2D3 (Fig. 5-6A), CYP7A1 mRNA (Fig. 5-6B) and protein (Fig. 5-6C).

Figure 5-6. Cholesterol is also correlated with 1,25(OH)2D3 and CYP7A1 expression in untreated human liver tissue. As in mice (Fig. 5-5), cholesterol was inversely correlated with (A) 1,25(OH)2D3 levels, (B) CYP7A1 mRNA, and (C) CYP7A1 protein expression in human liver tissue. Each data point represents one untreated human liver sample.

5.4.4 Studies 2 and 3: Elevated cholesterol caused by vitamin D deficiency and HF/HC diet

Mice maintained on the normal VD+ diet were switched over to the HF/HC VD+ diet during the last 3-weeks of the 8-week (Study 2) or 11-week (Study 3) diet periods. Since studies (8-weeks), conducted without or with Ca supplementation, showed that Ca supplementation did not alter

levels of 25(OH)D3, 1,25(OH)2D3, and VDR genes, the longer-term 11-week diet and intervention study was conducted without Ca supplementation. For most diets, the VD- mice displayed modest but significantly higher levels of plasma and liver cholesterol compared to those of normal VD+ controls. However, in mice fed the HF/HC VD- diet, vitamin D-deficiency exacerbated the elevation of cholesterol compared with mice fed the HF/HC VD+ diet (Figs. 5-7, A and B and 5- 8, A and B). Other common features could be identified: HF/HC VD+ controls displayed elevated

levels of cholesterol and total bile acid pool sizes (Figs. 5-7, A to C and 5-8, A to C), although the expression of all genes remained unchanged with the HF/HC diet, and the changes in Vdr and Fxr mRNA expression due to the VD- diet were small. However, Shp mRNA was significantly elevated (Fig. 5-7D) and Cyp7a1 mRNA was significantly decreased (Fig. 5-8D) with vitamin D- deficiency.

5.4.5 Study 2: Intervention with 1,25(OH)2D3

Following the short-term treatment of mice fed VD- diet (without Ca supplementation) with

1,25(OH)2D3 (2.5 µg/kg every other day for 1-week i.p.), plasma and liver cholesterol levels fell promptly to and below, respectively, levels of VD+ controls, indicating that treatment with

1,25(OH)2D3 could reduce cholesterol levels back to baseline. Accordingly, total bile acid pool sizes were increased with 1,25(OH)2D3 treatment (Fig. 5-7C), as expected of Cyp7a1 induction and higher cholesterol metabolism, although bile acid composition remained similar among the diets and treatment groups (Table 5-1). The concomitant attenuation of Shp mRNA triggered an elevation in Cyp7a1 mRNA and protein expression, especially for the HF/HC diet (Fig. 5-7D). In

contrast, ileal Fgf15 mRNA expression remained unchanged upon treatment with 1,25(OH)2D3

(Fig. A2-4A; Appendix II). These results were identical in a short-term 1,25(OH)2D3 intervention study that was performed in mice fed VD- diet with Ca supplementation (Fig. A2-2; Appendix II), suggesting that Ca in the diet is not a factor in the observed changes.

Table 5-1. Percent composition of total bile acid pool sizes in mice fed normal or combined high fat/high cholesterol VD+ or VD- diets for 8-weeks and treated with 1,25(OH)2D3 (2.5 µg/kg every other day during the last-week; Study 2) or for 11-weeks and treated 1 with D3 (20 µg/kg, every other day during the last 4-weeks; Study 3)

Study 2: Short-Term 1,25(OH)2D3 (0.47% Ca) Study 3: Long-Term Vitamin D3 (0.47% Ca) Normal Diet High Fat/High Cholesterol Diet Normal Diet High Fat/High Cholesterol Diet % composition VD+ Diet VD- Diet VD- Diet VD+ Diet VD- Diet VD- Diet VD+ Diet VD- Diet VD- Diet VD+ Diet VD- Diet VD- Diet

Control Control 1,25(OH)2D3 Control Control 1,25(OH)2D3 Control Control D3 Control Control D3

49±2.0 42±3.4 41±4.3 52±1.0a 49±1.6a 33±2.8b 63±3.4a 42±3.6b 48±8.1b 53±1.8# 54±1.5 54±2.5 t-CA

31±1.5a 40±3.7b 46±4.2b 35±1.1#a 34±1.7a 55±2.8b 24±2.3a 34±2.7b 37±9.5b 32±1.9# 32±1.5 34±2.7 t-βMCA

11±0.39 9.2±1.6 6.1±0.79 8.4±1.4a 10±1.1a 6.8±0.95b 7.7±0.57a 15±1.4b 9.2±0.66a 11±1.5a 10±0.48a 7.8±0.029b t-αMCA

7.5±0.52a 8.4±0.77a 4.4±0.93b 4.1±0.35# 5.4±0.79 4.1±0.27 5.4±1.0a 9.3±0.54b 6.1±2.5a 4.7±0.31a 3.9±0.23ab 3.6±0.23b t-ωMCA

1.7±1.7 0.0±0.0 2.1±1.3 0.29±0.29 0.35±0.35 1.2±0.58 0.15±0.065 0.13±0.0020 0.14±0.037 0.25±0.14 0.21±0.048 0.16±0.014 CA 1 Values are means ± SEMs, n=3-4 per group. Within each Study (2 or 3), labeled means within each diet (normal or high fat/high cholesterol) without a common letter differ for each bile acid, P<0.05. # difference between normal VD+ and high fat/high cholesterol VD+ controls in respective studies, P<0.05.

Figure 5-7. Short-term (1-week) intervention with 1,25(OH)2D3 (2.5 µg/kg every other day for 4 doses i.p.) decreases cholesterol in mice fed normal or high fat/high cholesterol (HF/HC) VD- diets for a total of 8-weeks. (A) Plasma and (B) liver cholesterol levels, which were elevated with the VD- diet were significantly reduced with 1,25(OH)2D3 treatment, whereas (C) total bile acid pool size (composed of t-CA, t-βMCA, t-αMCA, t-ωMCA, CA) was increased. (D) Treatment decreased Shp mRNA and increased Cyp7a1 mRNA and protein expression while liver Vdr and Fxr expression remained unchanged. Values are means ± SEMs, n = 3-6 per group; within respective diets (normal or HF/HC), labeled means without a common letter differ, P<0.05.

5.4.6 Study 3: Intervention with vitamin D3

When comparing plasma parameters of mice fed normal or HF/HC VD+ diet for a total of 11- weeks, normal VD+ diet controls displayed non-significantly lower levels of PTH (P=0.24), a finding that could be attributed to the slightly higher plasma Ca (P=0.22) that was also observed (Table 5-2). The VD- mice fed either a normal or HF/HC diet for a total of 11-weeks displayed

significantly (P<0.05) lower 25(OH)D3 and 1,25(OH)2D3 levels while PTH levels were increased compared to their respective VD+ diet controls; treatment with D3 readily reversed these effects

(Table 5-2). The treatment with D3 (20 µg/kg every other day for 4-weeks i.p.) to the D-deficient mice decreased plasma cholesterol in mice fed the HF/HC VD- diet (Fig. 5-8A), reduced liver cholesterol levels in D-deficient mice back to baseline (Fig. 5-8B), and increased total bile acid pool sizes (Fig. 5-8C) without changing the overall bile acid composition (Table 5-1). Treatment 89

reduced hepatic Shp mRNA expression (P<0.05) and increased Cyp7a1 mRNA expression (P<0.05), although the increase in Cyp7a1 protein expression was not significant (P=0.14 and P=0.49 for normal and HF/HC diets, respectively) (Fig. 5-8D). Similar to previous findings, treatment did not affect ileal Fgf15 mRNA expression (Fig. A2-4C; Appendix II). The shorter- term (1-week) treatment with D3 did not reduce cholesterol levels nor change expression of Shp and Cyp7a1 (data not shown).

Table 5-2. Plasma and liver parameters for mice fed the normal or high fat/high cholesterol VD+ or VD- diets for a total of 11- 1 weeks, with intraperitoneal treatment with 20 µg/kg D3 every other day during the last 4-weeks (Study 3)

Normal Diet High Fat/High Cholesterol Diet Parameters VD+ Diet VD- Diet VD- Diet VD+ Diet VD- Diet VD- Diet Control Control + D3 Control Control + D3 (vehicle-treated) (vehicle-treated) (vehicle-treated) (vehicle-treated) Plasma 25(OH)D3 (nM) 63.4 ± 1.77a 1.81 ± 0.663b 73.6 ± 1.22c 71.0 ± 1.83a 3.27 ± 1.01b 86.3 ± 3.51c Plasma 1,25(OH)2D3 (pM) 200 ± 35.7a 83.8 ± 15.0b 221 ± 23.0a 241 ± 10.9a 59.6 ± 4.96b 261 ± 4.90a Liver 1,25(OH)2D3 (pmol/kg tissue) 18.5 ± 2.60a 3.65 ± 1.54b 21.3 ± 3.74a 27.6 ± 10.6a 0.562 ± 0.611b 31.2 ± 4.10a Plasma Calcium (mg/dL) 12.9 ± 0.171 12.3 ± 0.323 12.2 ± 1.01 11.8 ± 0.797 11.7 ± 0.112 12.5 ± 0.554 Plasma Phosphorus (mg/dL) 36.7 ± 26.1a 184 ± 18.6b 60.2 ± 19.6a 74.7 ± 13.1a 313 ± 4.99b 106 ± 23.5a Plasma PTH (pg/mL) 63.4 ± 1.77a 1.81 ± 0.663b 73.6 ± 1.22c 71.0 ± 1.83a 3.27 ± 1.01b 86.3 ± 3.51c 1 Values are means ± SEMs, n = 4-6 per group. Within respective diets (normal or high fat/high cholesterol), labeled means without a common letter differ for each parameter, P<0.05

Figure 5-8. Long-term (4-weeks) intervention with D3 (20 µg/kg D3 every other day for 16 doses i.p.) decreases cholesterol in mice fed normal or high fat/high cholesterol (HF/HC) VD- diets for a total of 11-weeks. Vitamin D-deficiency (without Ca supplementation), together with the HF/HC diet, elevated (A) plasma and (B) liver cholesterol levels. Upon treatment with D3 for 4-weeks, (C) total bile acid pool size (composed of t-CA, t-βMCA, t-αMCA, t-ωMCA, CA) was increased, whereas (D) liver Vdr and Fxr mRNA expression remained unchanged and Vdr protein was increased in mice fed the HF/HC diet only. Treatment also decreased Shp mRNA expression, triggering increased Cyp7a1 mRNA expression in both normal and HF/HC diets. Values are means ± SEMs, n = 4-6 per group; within respective diets (normal or HF/HC), labeled means without a common letter differ, P<0.05.

5.5 Discussion

Recently, our laboratory established that, after repeated dosing of 1,25(OH)2D3, VDR-mediated suppression of Shp upregulated Cyp7a1 to lower cholesterol in mice (Chow et al., 2014). This novel mechanism is distinct from that of liver Fxr and intestinal Fgf15 normally known to regulate CYP7A1 for cholesterol lowering (Makishima et al., 1999; Goodwin et al., 2000; Lee et al., 2000; Chiang, 2004; Inagaki et al., 2005). The converse, that vitamin D-deficiency could result in hypercholesterolemia, however, remains untested. Hence, we investigated how vitamin D status affects Cyp7a1 and cholesterol homeostasis in mice. A vitamin D-deficient mouse model was first established, and the presence or absence of supplemented Ca in the VD- diet was found devoid of

effects on levels of 25(OH)D3, 1,25(OH)2D3, Ca (Figs. 5-2, A to D), or other VDR targets after 8- weeks of diet, as found in other studies (Vieth et al., 1987; Goff et al., 1992; Song and Fleet, 2007).

Seemingly, the low 1,25(OH)2D3 but high PTH levels counteracted each other to minimize changes in serum Ca. Similar to previous findings in rats (Goff et al., 1992), small increases in plasma

1,25(OH)2D3 levels were observed at 4-weeks in mice fed the VD- diet without Ca

supplementation, but these discrepancies disappeared by 8-weeks (Fig. 5-2B). Both 25(OH)D3 and

1,25(OH)2D3 levels in the VD- mice were decreased to 13-30% of VD+ controls after 8-weeks of diet. A similar vitamin D-deficient model was established in the rat after 12-weeks of diet (Li et al., 2016).

With this vitamin D-deficient model, we proceeded to test the corollary that vitamin D-deficiency could alter downstream Vdr targets, including hepatic Shp and Cyp7a1 expression in cholesterol

homeostasis. We found critical and significant correlations among 1,25(OH)2D3, cholesterol, Vdr, Shp, and Cyp7a1 expression (Figs. 5-4 and 5-5) in the mouse liver. The data revealed that, with

reduced Vdr expression and lower 1,25(OH)2D3 levels in vitamin D-deficiency, Shp was elevated, leading to reduced Cyp7a1 mRNA and protein expression and elevated cholesterol levels downstream. Negative correlations were found with cholesterol and 1,25(OH)2D3 or Cyp7a1 mRNA or protein expression in mouse livers (Fig. 5-5). This observation showed, for the first time, a direct link between vitamin D status and hypercholesterolemia. Interestingly, similar

negative correlations, albeit non-significant, were found between cholesterol and 1,25(OH)2D3

levels, as well as cholesterol and CYP7A1 expression in human livers (Fig. 5-6), suggesting that the relationship between vitamin D-deficiency and hypercholesterolemia could also exist in humans. Further investigation is required to dissect confounding variables that may influence these correlations, including the consideration of age, sex, and status of health.

For further perturbation of cholesterol levels in the D-deficient model, we combined the VD- model with HF/HC diets. Plasma PTH levels were higher with HF/HC diets, albeit non- significantly, possibly due to a protective role in which excess PTH stimulates the breakdown of fat (Bousquet-Melou et al., 1995) or merely a homeostatic result of the lower plasma Ca levels that were observed. For mice fed the HF/HC diets, plasma and liver cholesterol levels were 1.6- and 3.7-fold those of normal diet controls, and levels were further exacerbated with vitamin D- deficiency. Consistent with previous findings (Chow et al., 2014), 1,25(OH)2D3 treatment at 93

pharmacological doses (2.5 µg/kg) was effective in both normal and hypercholesterolemic VD- mice and promptly increased Cyp7a1 protein expression via downregulation of hepatic Shp to decrease plasma and liver cholesterol, events that led to increased bile acid pool sizes (Fig. 5-7). Interestingly, a longer treatment of VD- mice with physiological doses of the inactive precursor,

D3, also triggered downregulation of Shp and induction of Cyp7a1 to reduce cholesterol levels

back to baseline. The effectiveness of D3 was attributed to its ability to increase levels of 25(OH)D3 and 1,25(OH)2D3 back to those of VD+ mice (Table 5-2), but required a longer treatment regimen compared with 1,25(OH)2D3. However, the therapeutic potential of D3 in cholesterol lowering is particularly intriguing as it acts without inducing hypercalcemia (Table 5-2), a toxic effect that

limits the therapeutic use of 1,25(OH)2D3.

Much speculation exists regarding the relationship between vitamin D status and hypercholesterolemia (Karhapaa et al., 2010; Schnatz et al., 2014). Presently, we have provided evidence to suggest that vitamin D- deficiency is directly correlated with hypercholesterolemia via altered Vdr, Shp, and Cyp7a1 expression. Previously, Li et al. (2016) proposed a relationship in rats where deficiency reduced Vdr expression and increased cholesterol synthesis via induction of Srebp2 and Hmgcr. When we examined the effects of vitamin D-deficiency on the cholesterol synthesis pathway, we observed only a minor, non-significant (P=0.42) increase in Hmgcr mRNA expression in VD- compared with VD+ mice, and these levels were neither correlated with liver

1,25(OH)2D3 nor cholesterol levels (Fig. A2-3; Appendix II). Furthermore, Srebp2 mRNA expression was unchanged (Fig. A2-3; Appendix II). Another potential mechanism to explain the relationship lies within the common synthesis pathway precursor, 7-dehydrocholesterol (7-DHC), which is either converted to cholesterol via 7-dehydrocholesterol reductase (DHCR7) or reacts

with UV-B light to produce vitamin D3 in skin. Thus, the balance between conversion of 7-DHC

to cholesterol or vitamin D3 can be modulated by altering either DHCR7 levels or exposure to UV- B light. In rats when DHCR7 activity was inhibited, there was a marked reduction in plasma cholesterol levels and enormous increase in 7-DHC concentration (Xu et al., 1995), hallmarks of Smith-Lemli-Opitz syndrome, where inhibition of DHCR7 switches the flux of 7-DHC from

cholesterol to vitamin D3 synthesis (Prabhu et al., 2016). Insufficient sunlight exposure is associated with both vitamin D-deficiency (Mithal et al., 2009) and increased serum cholesterol (Grimes et al., 1996). The lack of UV-B light from housing VD- mice in the dark may have

contributed to the switch from formation of vitamin D3 to cholesterol, albeit, vitamin D3 synthesis 94

in the skin of rodents was previously reported to be minor and further hampered by the presence of hair (Lawson et al., 1986). We observed a slight increase (P=0.06) in liver Dhcr7 mRNA expression in VD- mice; however, the levels were once again neither correlated with liver

1,25(OH)2D3 nor cholesterol (Fig. A2-3; Appendix II). Hence, the evidence unveiled thus far suggests that the hypercholesterolemia observed in VD- mice is primarily attributed to the altered expression of cholesterol-regulatory genes in the metabolism pathway and not the synthetic pathway.

Collectively, we have provided compelling evidence that liver 1,25(OH)2D3 is positively associated with Vdr and Cyp7a1 expression and inversely correlated with Shp expression in mice. Interestingly, both mouse and human liver cholesterol levels were found to be inversely correlated

with 1,25(OH)2D3 and Cyp7a1/CYP7A1 expression. Hence, the key regulators of the cholesterol metabolism pathway appear to be altered by vitamin D-deficiency and contribute to hypercholesterolemia, observations that are reversed upon replenishment of vitamin D. Although intervention with 1,25(OH)2D3 and D3 suppressed Shp to increase Cyp7a1, no apparent change was observed with sterol 12α-hydroxylase or Cyp8b1 (Figs. A2-5 and A2-8; Appendix II), another target of Shp-mediated repression that controls the relative amounts of CDCA to CA (Li-Hawkins et al., 2002). The observation was further supported by the lack of significant change in the bile acid composition. Collectively, we found no evidence to suggest that the regulation of Cyp8b1 is affected by altered Vdr in vitamin D-deficiency, liver 1,25(OH)2D3, or Shp expression (data not shown). Other factors that influence cholesterol homeostasis were also examined, namely mediators of cholesterol uptake (LDL receptor, scavenger receptor class B member 1, and Niemann-Pick C1 Like 1 transporter) and cholesterol efflux (ABCA1, ABCG5, and ABCG8

transporters) (van der Wulp et al., 2013). While treatment of VD- mice with 1,25(OH)2D3 slightly decreased the hepatic but not ileal expression of Abca1, Abcg5, and Abcg8, no other trend was observed with VD- diet or intervention (Figs. A2-5 and A2-6; Appendix II). Nonetheless, it remains important to emphasize that maintaining sufficient levels of vitamin D is critical in the prevention of hypercholesterolemia to lower the risk for developing cardiovascular disease. For future studies, LDL receptor knockout mice could be utilized to clarify the mechanism of vitamin D-deficiency-associated hypercholesterolemia

5.6 Acknowledgments

This work was supported by the Canadian Institutes of Health Research (KSP), the National Sciences and Engineering Research Council of Canada (HPQ) and the Ontario Scholarship Program (HPQ). We thank Dr. Matthew R. Durk for his assistance in designing the diets.

5.7 Supplementary material

Supplementary Table S5-1. Donor information for human liver samples (untreated) used for correlations Age Height Weight Body Mass Sample ID Gendera Raceb Blood Type Cause of Deathc (years) (inches) (kg) Index HH1016 F H 64 NA NA NA NA CVA-ICH HH1028 M H 43 65 65 23.87 O+ CVA-ICH HH1041 M C 53 70 181 57.12 A CVA second to ICH HH1047 M H 44 67 68 23.52 A+ HT second to blunt injury HH1057 F C 33 65 57 21.2 O+ CVA stroke HH1072 F C 40 67 108 37.3 A+ CNS tumor second to stroke HH1078 F C 53 61 68 28.3 A CVA second to ICH HH1076 M C 34 69 110 35.58 A+ HT second to blunt injury, MVA HH1089 F C 57 66 80 28.4 A- CVA/stroke HH1065 F C 66 66 44 16.1 O+ CVA/stroke HH1071 M H 18 70 84 26.19 O+ HT second to blunt injury NA not available a Female (F), Male (M) b Hispanic (H), Caucasian (C) c cerebral vascular accident (CVA), intracerebral hemorrhage (ICH), central nervous system (CNS), head trauma (HT), motor vehicle accident (MVA)

Supplementary Table S5-2. Composition of normal and high fat/high cholesterol VD+ and VD- diets. These diets were used in establishing the vitamin D-deficiency model (Study 1) and intervention with 1,25(OH)2D3 (Study 2) or D3 (Study 3) Normal Diet High Fat/High Cholesterol Diet Vitamin D- Vitamin D- Vitamin D- Vitamin D- Vitamin D- Vitamin D- Sufficient Deficient Deficient Sufficient Deficient Deficient TD.07370 TD.89123 TD.07541 TD.130146 TD.140244 TD.130148 Fat (%) 0 0 0 44 42 44 Cholesterol (%) 0 0 0 0.2 0.2 0.2 Vitamin D3 (IU/g) 2.2 0 0 2.2 0 0 Calcium (%) 0.47 0.47 2.5 0.7 0.7 2.5 Phosphorus (%) 0.3 0.3 1.5 0.5 0.5 1.5

Supplementary Table S5-3. Bile acid standards and parameters for LC-MS/MS for measuring bile acid pool sizes of mice a following intervention with 1,25(OH)2D3 (Study 2) or D3 (Study 3) Internal Standard Declustering potential Collision energy Cell exist potential Name Symbol m/z (IS) (V) (eV) (V)

Cholic Acid CA 407.3→407.3 CA-d4 -125 -25 -5 b Chenodeoxycholic acid (as sodium salt) CDCA 391.3→391.3 CDCA-d4 -130 -25 -5 b Deoxycholic acid DCA 391.3→391.3 DCA-d4 -130 -25 -5 b Lithocholic acid LCA 375.3→375.3 LCA-d4 -120 -23 -5 Taurocholic acid (as sodium salt hydrate) t-CA 514.3→79.9 CA-d4 -120 -120 -5 Tauro-α-muricholic acid (as sodium salt) t-αMCA 514.3→79.9 CA-d4 -120 -120 -5 Tauro-β-muricholic acid (as sodium salt) t-βMCA 514.3→79.9 CA-d4 -120 -120 -5 Tauro-ω-muricholic acid (as sodium salt) t-ωMCA 514.3→79.9 CA-d4 -120 -120 -5 Cholic-2,2,-4-4-d4 Acid CA-d4 411.3→411.3 -125 -25 -5 Deoxycholic -2,2,-4-4-d4 Acid DCA-d4 395.3→395.3 -130 -25 -5 Chenodeoxycholic -2,2,-4-4-d4 Acid CDCA-d4 395.3→395.3 -130 -25 -5 Lithocholic-2,2,-4-4-d4 Acid LCA-d4 379.3→379.3 -120 -23 -5 a CDCA, CA, DCA, LCA, and t-CA were obtained from Sigma-Aldrich (Mississauga, ON); t-αMCA, t-βMCA, and t-ωMCA were from Steraloids (Newport, RI); CA-d4, DCA-d4, CDCA-d4, and LCA-d4 were from CDN Isotopes (Pointe-Claire, QC) bCDCA, DCA, and LCA were undetectable.

Supplementary Figure S5-1. Basal 1,25(OH)2D3 concentrations and plasma PTH and renal Cyp27b1 and Cyp24a1 expression in mice fed normal VD+ or VD- diets. At the end of the 8- week diet, changes for the VD- diets, without or with Ca, were similar and thus data were used collectively under the same VD- classification. (A) Plasma, liver, and kidney 1,25(OH)2D3 concentrations and (B) tissue-to-plasma partitioning ratios were measured and calculated. Values are means ± SEMs (n = 12-20). (C) When mice were fed VD+ (white) or VD- diets without (grey) or with (black) Ca supplementation for 0- to 8-weeks, plasma PTH was non-significantly increased with VD- diets at all weeks. Concurrently, renal (D) Cyp27b1 and (E) Cyp24a1 mRNA expression were significantly increased and decreased, respectively, after 8-weeks of VD- diets. Values are means ± SEMs (n = 3-7); labeled means at a time without a common letter differ, P<0.05.

5.8 Statement of significance of Chapter 5

Upon examination of the literature, numerous clinical studies have speculated an association between vitamin D-deficiency and hypercholesterolemia. The observational nature of these studies, however, has limited the ability to draw conclusions on causality and potential molecular mechanisms that could link the two together. Seemingly, the balance between cholesterol synthesis and metabolism determines the levels of cholesterol in the body. As such, hypercholesterolemia develops because of either increased synthesis, decreased metabolism, or a combination of the two. Recently, Li et al. (2016) suggested that vitamin D-deficiency-induced hypercholesterolemia 99

in rats was due to enhanced synthesis via Hmgcr rather than reduced catabolism via Cyp7a1. Under vitamin D-deficient conditions, Vdr activity was reduced, decreasing the repressive role of insulin- induced gene-2 (Insig-2) expression on Srebp2 and Hmgcr (Li et al., 2016). However, the role of altered Cyp7a1 due to vitamin D-deficiency cannot be ruled out based on this report alone. Previously, our laboratory reported that the expression of Cyp7a1 in the rat is affected by very low levels of hepatic Vdr, as well as the presence of secondary hepatic Fxr effects that occurred as a result of Vdr-mediated induction of intestinal Asbt to increase bile acid absorption into the portal blood (Chow et al., 2009).

In the previous chapter, we demonstrated that the Vdr plays a direct role in cholesterol regulation by inhibition of Shp to increase Cyp7a1 expression and cholesterol metabolism in mice. Here, we now show that the corollary exists. Namely, vitamin D deficiency caused reduced Vdr expression, upregulation of Shp, inhibition of Cyp7a1, and elevated cholesterol levels. Notably, we found significant positive correlations between Vdr and Cyp7a1 vs. liver 1,25(OH)2D3 levels, and inverse

correlations between Shp vs. liver 1,25(OH)2D3 and Cyp7a1 vs. Shp. These findings suggest that the expression of these cholesterol-regulating genes is directly related to vitamin D status.

Furthermore, significant inverse correlations exist between cholesterol and 1,25(OH)2D3 levels in the liver, as well as cholesterol vs. Cyp7a1. Interestingly, human liver cholesterol levels were also inversely correlated with 1,25(OH)2D3 and CYP7A1, indicating that a similar mechanism of vitamin D deficiency-induced hypercholesterolemia could exist in humans and could be used to explain the commonly reported association between vitamin D deficiency and hypercholesterolemia in clinical studies.

We also demonstrated that intervention of vitamin D-deficient mice with 1,25(OH)2D3 and/or

vitamin D3 could reverse the altered expression of cholesterol-regulating genes that was caused by deficiency. In summary, we have identified the VDR as a novel therapeutic target for

hypercholesterolemia, since the administration of 1,25(OH)2D3 and vitamin D3 was able to inhibit Shp and increase Cyp7a1 expression to reduce elevated cholesterol levels back to baseline.

My contributions to this chapter include design of experiments, validation of vitamin D deficiency

models, maintenance and treatment of mice, tissue harvesting, determination of plasma 25(OH)D3,

1,25(OH)2D3, calcium, PTH, and cholesterol, extractions and determination of liver cholesterol

100

and 1,25(OH)2D3, extractions for measurement of bile acid pool size by LC-MS/MS, mRNA and protein analyses, and writing the paper. Stacie Y. Hoi measured mRNA expression of genes in the model validation experiments in vitamin D-deficient mice supplemented with calcium. Jie Chen contributed to the maintenance and treatment of mice for the long-term vitamin D3 and bile acid pool size studies and measured mRNA expression. Adrie Bruinsma measured the mRNA expression in the kidney and intestine for the vitamin D-deficient mice supplemented with calcium. Geny M.M. Groothuis provided comments and edited the paper. Albert P. Li provided the human liver tissues. Edwin C. Y. Chow developed the LC-MS/MS method and provided valuable discussion. K. Sandy Pang contributed to design of experiments, analyses of data, and writing the paper.

101

Chapter 6

Vitamin D Analogs for Cholesterol Lowering

Holly P. Quach, Paola Bukuroshi, Tamara Dzekic, Lilia Magomedova, Carolyn L. Cummins, and K. Sandy Pang

Department of Pharmaceutical Sciences, Leslie Dan Faculty of Pharmacy, University of Toronto, Toronto, Ontario, Canada

102

6.1 Abstract

1α,25-Dihydroxyvitamin D3 [1,25(OH)2D3], active ligand of the vitamin D receptor (VDR), was shown to reduce cholesterol via induction of the cholesterol metabolizing enzyme, cholesterol 7α-hydroxylase (Cyp7a1), upon down-regulation of the small heterodimer partner (Shp), which represses Cyp7a1, in mice. Due to the high propensity of

1,25(OH)2D3 for hypercalcemia, we appraised other vitamin D analogs: vitamin D3, 25-

hydroxyvitamin D3 [25(OH)D3], 1α-hydroxyvitamin D3 [1α(OH)D3], and 1α-

hydroxyvitamin D2 [1α(OH)D2] as alternate VDR ligands. In vitro potency assays with transfected HEK293 cells (GAL4-hVDR, UAS-luciferase reporter and β-galactosidase) and the determination of VDR-mediated induction of transporter and enzyme expression in Caco-2 cells provided conflicting results. In transfected HEK293 cells, higher potencies

(lower EC50) for 1α(OH)D3 and 25(OH)D3 (302 vs. 307 nM) were observed compared with

1α(OH)D2 and vitamin D3 (650 and 2130 nM). But in Caco-2 cells, greater 1α(OH)D3- than 25(OH)D3-related induction of VDR targets was observed. These results were explained by the high CYP27B1 expression in HEK293 cells and CYP2R1 expression in

Caco-2 cells that are involved in the bioactivation of 25(OH)D3 and 1α(OH)D3, respectively, to 1,25(OH)2D3. Upon addition of ketoconazole (KTZ), a P450 inhibitor, the

apparent potency of 25(OH)D3, expressed as EC50, increased from 200 to 608 nM in HEK293 cells, and there was also a general blunting of VDR target gene expression in Caco-2 cells. In C57BL/6 mice that exhibited elevated plasma and liver cholesterol levels after a 3-week Western diet, treatment with vitamin D analogs intraperitoneally during the last week of diet revealed a rank order for cholesterol lowering. Cholesterol reduction was accompanied by a reduction in hepatic Shp and induction of Cyp7a1 expression upon

vitamin D analog treatment: 1.75 nmol/kg 1α(OH)D3 > 1248 nmol/kg 25(OH)D3 >> 1625 nmol/kg vitamin D3 and 1.21 nmol/kg 1α(OH)D2, with the latter two analogs failing to

lower cholesterol. A greater in vivo potency was suggested for 1α(OH)D3 since the dose

ratio of 1α(OH)D3:25(OH)D3 was very low (0.0014). The in vitro and in vivo results for the vitamin D analogs are explained by differential basal levels of enzymes in HEK293 and Caco-2 cells, emphasizing the need to consider enzyme bioactivation when interpreting potencies of vitamin D analogs.

103

6.2 Introduction

1α,25-Dihydroxyvitamin D3 [1,25(OH)2D3], the natural and active ligand of the vitamin D

receptor (VDR), is formed from the sequential bioactivation of vitamin D3 to 25-

hydroxyvitamin D3 [25(OH)D3] via 25-hydroxylases (CYP2R1, CYP27A1) in the liver,

followed by conversion to 1,25(OH)2D3 via 1α-hydroxylase (CYP27B1) in the kidney. For

homeostasis, 1,25(OH)2D3 is rapidly degraded by 24-hydroxylase (CYP24A1), a VDR target enzyme, followed by a series of metabolic pathways to form calcitroic acid for excretion (Jones et al., 1998). Although the traditional role of the VDR is to regulate plasma calcium (via the absorptive calcium channels TRPV5 and 6) (den Dekker et al., 2003) and phosphate levels (Jones et al., 1998), recent studies have revealed that the VDR is also an important regulator of transporters and enzymes involved in human drug disposition and metabolism, including P-glycoprotein (P-gp) (Chow et al., 2011a), organic anion- transporting polypeptide 1A2 (OATP1A2) (Eloranta et al., 2012), sulfotransferase 2A1 (SULT2A1) (Echchgadda et al., 2004), and cytochrome P450 3A4 (CYP3A4) (Schmiedlin-Ren et al., 2001).

The lipophilic nature of 1,25(OH)2D3 allows for rapid equilibration between plasma and tissue sites to activate the VDR to exert its traditional calcium-regulating actions in the kidney, intestine, and bone in mice in vivo (Chow et al., 2013b). Due to its lipophilicity,

1,25(OH)2D3 in plasma also rapidly equilibrates with that in mouse liver to activate VDR for down-regulation of the small heterodimer partner (Shp) and induction of cholesterol 7α-hydroxylase (Cyp7a1) (Chow et al., 2014), the rate-limiting enzyme for cholesterol metabolism to bile acids (Chiang, 2004). The primary negative feedback mechanism for CYP7A1 is the farnesoid X receptor (FXR) and SHP regulatory cascade (Goodwin et al., 2000), while a second mechanism exists in the intestine via activation of FXR to induce fibroblast growth factor 15/19 (rodent/human Fgf15/FGF19), a hormonal signaling molecule that represses hepatic CYP7A1 through interaction with the hepatic fibroblast growth factor receptor 4 (Inagaki et al., 2005). 1,25(OH)2D3-liganded VDR has been shown to antagonize FXR in vitro (Honjo et al., 2006) and induce Fgf15 expression in vivo (Schmidt et al., 2010). In rat livers where Vdr levels are extremely low (Gascon-Barre et al., 2003), changes in Cyp7a1 expression are insignificant (Chow et al., 2009; Chow et al., 104

2011b). In contrast, in mouse livers that contain measureable Vdr protein expression

(Gascon-Barre et al., 2003; Chow et al., 2014), treatment with 1,25(OH)2D3 led to increased mRNA and protein expression of hepatic Cyp7a1 (Chow et al., 2014). Thus,

1,25(OH)2D3-mediated activation of the VDR could lead to cholesterol lowering and act

as a potential therapeutic target. However, the utility of 1,25(OH)2D3 is limited because of its tendency to elicit hypercalcemia. Alternatively, vitamin D analogs that do not exhibit hypercalcemic effects are desired as therapeutic agents.

Vitamin D3, 25(OH)D3, and 1α-hydroxyvitamin D3 [1α(OH)D3] are precursors that

undergo metabolic conversion to the active ligand, 1,25(OH)2D3. In tissues where VDR is

present, VDR binding affinity is dependent on structural features, with 1,25(OH)2D3 >

1α(OH)D3 > 25(OH)D3 >> vitamin D3 (Eisman and DeLuca, 1977). Although the binding

affinity of vitamin D3 to the VDR is very low, treatment with vitamin D3 could trigger VDR

effects upon bioactivation, through increasing serum levels of 25(OH)D3 or 1,25(OH)2D3 in rodents (Fleet et al., 2008; DeLuca et al., 2010) and humans (Bischoff-Ferrari et al., 2012; Navarro-Valverde et al., 2016).

The VDR activity of 25(OH)D3 is thought to be mediated through 1,25(OH)2D3, although

25(OH)D3 is known to directly bind and activate the VDR, albeit at a much lower affinity

than 1,25(OH)2D3 (Eisman and DeLuca, 1977; DeLuca et al., 2010). In rats in vivo,

escalating doses of 25(OH)D3 triggered progressively decreased renal Cyp27b1 activity but increased activity of renal Cyp24a1, respectively, without increasing serum

1,25(OH)2D3 levels (Vieth et al., 1990b). In both wildtype and Cyp27b1 knockout mice

where 25(OH)D3 could not be converted to 1,25(OH)2D3, treatment with 25(OH)D3 increased serum calcium levels and caused toxicity, suggesting that 25(OH)D3 itself may indeed be biologically active (DeLuca et al., 2010). Moreover, the incubation of Caco-2

cells with 25(OH)D3 has led to increased protein expression of P-gp and multidrug resistance associated protein 2 (MRP2) (Fan et al., 2009).

In addition to direct binding to the VDR, 1α(OH)D3, although not an endogenous substrate,

is continuously being converted to 1,25(OH)2D3 by enzymes such as CYP2R1 (Jones,

2008). Due to its greater lipophilicity, 1α(OH)D3 exerts a greater effect on VDR target

105

genes compared with 1,25(OH)2D3 at equimolar doses, giving rise to a higher extent of

hypercalcemia than observed with 1,25(OH)2D3 (Chow et al., 2013a). Previous studies

have shown that 1α(OH)D3 increased VDR target genes such as renal Cyp24a1, Mrp2, Mrp3, and Mrp4 in mice in vivo (Nishida et al., 2009), increased protein expression of P- gp, MRP2, MRP4, and CYP3A4 in Caco-2 cells (Fan et al., 2009), decreased CYP7A1 expression in human hepatocytes (Han and Chiang, 2009), or increased Cyp7a1 expression

in mouse livers (Nishida et al., 2009), thereby suggesting a potential role of 1α(OH)D3 in cholesterol lowering.

Doxercalciferol [1α-hydroxyvitamin D2 or 1α(OH)D2] is another vitamin D analog that is metabolized by 25-hydroxylase and 24-hydroxylase to form the active metabolites,

1,25(OH)2D2 and 1,24(OH)2D2, respectively (St Peter, 2000), and is used for the treatment of secondary hyperparathyroidism (Frazao et al., 2000), metabolic bone disease (Strugnell et al., 1995), and tumor growth (Grostern et al., 2002). Recently, it was shown that

treatment with 1α(OH)D2 could activate the VDR to decrease the accumulation of triglycerides and cholesterol in the mouse kidney without triggering hypercalcemia (Wang

et al., 2011). Comparison of the effects of 1α(OH)D2 with that of 1,25(OH)2D3 in the Caco-

2 cell monolayer revealed that 1α(OH)D2 was equipotent to 1,25(OH)2D3 for induction of mRNA expression of multidrug resistance gene 1 (MDR1), MRP2, CYP3A4, and protein expression of P-gp and MRP4 (Fan et al., 2009).

In this communication, we examined the in vitro potencies and transcriptional activities of

various analogs of 1,25(OH)2D3 in the presence or absence of ketoconazole (KTZ), a general inhibitor of cytochrome P450s that would decrease the conversion of the analogs

to active 1,25(OH)2D3, and further tested their potential as non-toxic cholesterol lowering agents in mice in vivo.

106

6.3 Materials and methods

6.3.1 Materials

Vitamin D3, 25(OH)D3, 1α(OH)D3, and 1,25(OH)2D3 powders were procured from Sigma-

Aldrich Canada (Mississauga, ON). 1α(OH)D2 was a kind gift from Genzyme (Cambridge, MA). Antibodies to Cyp7a1 (N-17) (Santa Cruz Biotechnology, Santa Cruz, CA), CYP27B1 (C-terminus) and Gapdh (6C5) (Abcam, Cambridge, MA) were purchased for

the study. The enzyme-immunoassay kit for 1,25(OH)2D3 measurement was manufactured by Immunodiagnostics Systems Inc. (Scottsdale, AZ) and obtained from Inter Medico (Markham, ON). The Pierce LDH cytotoxicity assay kit, Dulbecco’s modified Eagle’s medium (DMEM), fetal bovine serum (FBS), nonessential amino acids, and penicillin- streptomycin were all purchased from Life Technologies (Burlington, ON) and D-luciferin sodium salt was purchased from Invitrogen (Burlington, ON). The plasmids used in the luciferase reporter assay were a kind gift from Dr. David J. Mangelsdorf (University of Texas Southwestern Medical Center, Dallas, TX). Human embryonic kidney (HEK293) and human colon (Caco-2) cells were obtained from the American Type Culture Collection (Manassas, VA). All other reagents were purchased from Sigma-Aldrich Canada or Fisher Scientific (Mississauga, ON).

6.3.2 Luciferase reporter assay for in vitro binding to VDR

HEK293 cells were seeded at a density of 30000 cells/well in 96-well plates and maintained in DMEM supplemented with 10% FBS. Transfection assays were performed in media containing 10% charcoal-stripped FBS using calcium phosphate. The total amount of plasmid DNA (150 ng/well) included 50 ng UAS-luc reporter, 20 ng β-galactosidase, 15 ng GAL4-hVDR fusion protein and pGEM filler plasmid. Ligands were added at 6 to 8 h post-transfection. Cells were harvested 14 to 16 h later were then assayed for luciferase and β-galactosidase activity. Relative luciferase units (RLU) were calculated as: (luciferase light units/β-galactosidase)*time. For investigation of the bioactivation of vitamin D analogs to 1,25(OH)2D3, 10 µM KTZ was added and co-incubated with 0.1% ethanol

(control), 1,25(OH)2D3, 25(OH)D3, or 1α(OH)D3. The EC50 was estimated from the curve

107

(Emin -E max ) model, after fitting data to a 4-parameter logistic model RLU = γ +Emax with 1+(C/EC50 ) GraphPad Prism Software (version 6; GraphPad Software Inc., La Jolla, CA).

6.3.3 In vitro potencies of 25(OH)D3, 1α(OH)D3, and 1,25(OH)2D3 in HEK293 and Caco-2 cells, in presence or absence of ketoconazole

Since the vitamin D analogs are prone to metabolism or bioactivation in the presence of enzymes such as CYP2R1 and CYP27B1, KTZ was used as an inhibitor to re-assess the potencies of the vitamin D analogs tested. HEK293 cells were grown in 10-cm plates in DMEM supplemented with 10% FBS and 1% penicillin-streptomycin under an atmosphere

of 5% CO2 and 95% relative humidity at 37°C. At 85% confluency, HEK293 cells were seeded at a density of 30000 cells/well in 96-well plates and then treated for 24 h with 0.1%

ethanol, 100 nM 25(OH)D3, or 100 nM 1α(OH)D3 in 0.1% ethanol without or with 10 µM

KTZ to assess the role of CYPs in the conversion of analogs to 1,25(OH)2D3. At the end of the 24 h incubation period, cells were harvested for protein measurement. The experiments were repeated three times, with sampling conducted in triplicate.

Caco-2 cells were cultured in DMEM supplemented with 10% FBS, 1% nonessential amino acids, 100 U/ml penicillin, and 100 µg/ml streptomycin under an atmosphere of 5%

CO2 and 95% relative humidity at 37°C. All experiments were performed with cells at passage numbers 18 to 24. Caco-2 cells were seeded at a density of 25000 cells/cm2 in 60- mm dishes and the medium was changed every other day, with the exception of treatment days, as previously described (Fan et al., 2009). Cells were treated with control medium containing 0.1% ethanol, 100 nM 25(OH)D3, 100 nM 1α(OH)D3, or 100 nM 1,25(OH)2D3, in 0.1% ethanol, without or with 10 µM KTZ daily, on day 18 for 3 consecutive days. Cells were harvested on day 21 for mRNA or protein measurement. Experiments were performed three times with sampling conducted in triplicate. Dose-response and LDH cytotoxicity studies with KTZ (0-25 µM) were also performed and no toxicity was found (data not shown).

108

6.3.4 In vivo potencies of vitamin D analogs in mice fed a Western diet

Male C57BL/6 mice (8 weeks old; n = 3-7) were maintained under a 12:12-h light and dark cycle in accordance with approved protocols by the Animal Care and Use Committee at the University of Toronto. Mice were fed a normal or Western diet (WD; TD.88137) prepared by Harlan Laboratories (Madison, WI) containing high fat (42%) and high cholesterol (0.2%) for a total of 3 weeks, as in previous studies, to elevate plasma and liver cholesterol levels (Chow et al., 2014). At the beginning of the last week of the WD, mice were given vitamin D3 (1625 nmol/kg), 25(OH)D3 (1248 nmol/kg), 1α(OH)D3 (1.75 nmol/kg), or 1α(OH)D2 (1.21 nmol/kg) by intraperitoneal (i.p.) injection at 10 a.m. every other day for 8 days. Systemic and portal blood samples and tissues were obtained under anesthesia with i.p. administration of ketamine and xylazine (150 and 10 mg/kg, respectively) at 50 h after the last dose at 12 p.m. The depth of anesthesia was assessed by monitoring the heart rate and pedal reflex. The abdominal vena cava was flushed with ice- cold saline, before harvesting of the liver, kidney and scraped enterocytes (ileum) tissues as described (Chow et al., 2011a).

6.3.5 Plasma analysis

Plasma calcium and phosphorus from systemic blood were quantified by inductively coupled plasma atomic emission spectroscopy (Optima 3000 DV, Perkin Elmer Canada, Woodbridge, ON) as described by Chow et al. (2011a). Alanine aminotransferase (ALT) concentrations in systemic blood were determined using an ALT Reagent kit (BioQuant Diagnostics Inc., San Diego, CA) and total bile acid levels in the portal blood were quantified using the Total Bile Acid kit (Diazyme Laboratories, Poway, CA).

6.3.6 Determination of cholesterol concentrations in plasma and liver

Total plasma cholesterol levels were determined by the Total Cholesterol Kit (Wako Diagnostics, Richmond, VA). For liver cholesterol measurements, lipids were extracted from ~0.2 g liver tissue homogenized in chloroform:methanol (2:1 v/v), using the Folch method (Folch et al., 1957). Cholesterol concentrations were determined from extracts using Infinity Cholesterol reagents (Thermo Scientific, Rockford, IL). 109

6.3.7 Determination of 1,25(OH)2D3 concentrations in plasma and liver

Liver samples were extracted as previously described (Chow et al., 2013b; Chow et al.,

2014). Both plasma and liver samples were delipidated and 1,25(OH)2D3 concentrations were then assayed by EIA as per manufacturer’s protocol.

6.3.8 Relative mRNA expression using quantitative PCR

Total RNA extraction and cDNA synthesis procedures were identical to those previously described (Chow et al., 2011a). The primer sequences are summarized in Supplementary

Table S6-1. The critical threshold cycle (CT) values of target genes for mouse liver and kidney were normalized to Cyclophilin, whereas intestinal samples were normalized to Villin, and HEK293 and Caco-2 cells were normalized to GAPDH.

6.3.9 Quantification of relative protein expression by Western blotting

Protein isolation was performed as previously described (Chow et al., 2010). Protein samples (50 μg) were loaded and separated by 10% SDS-polyacrylamide gels, then transferred onto nitrocellulose membranes (Amersham Biosciences, Piscataway, NJ). Cyp7a1, CYP27B1, and CYP2R1 protein expression was determined by Western blotting, as previously described (Fan et al., 2009; Chow et al., 2010). Cyp7a1, CYP27B1, and CYP2R1 protein expression was normalized to Gapdh/GAPDH.

6.3.10 Statistics

For in vitro studies, data are expressed as mean ± SD. For in vivo studies, data are expressed as mean ± SEM. For comparison of data between two groups, the Student’s t-test was used and P < 0.05 was set as the level of significance.

110

6.4 Results

6.4.1 Vitamin D analogs are less transcriptionally active than

1,25(OH)2D3 in vitro

The GAL4-hVDR luciferase reporter assay was used to determine the transcriptional activities of the vitamin D analogs in HEK293 cells. All analogs were found to be less

active compared to 1,25(OH)2D3, which displayed an EC50 of 2.6 nM in absence of KTZ.

The apparent EC50 of 1α(OH)D3 was similar to that for 25(OH)D3 (302 vs. 307 nM), but

was higher for 1α(OH)D2 (650 nM) and vitamin D3 (2130 nM) (Fig. 6-1A). Upon addition

of 10 µM KTZ, the luciferase activity following treatment with 25(OH)D3 was significantly blunted (EC50 increased from 200 to 608 nM), while activities of 1,25(OH)2D3 and 1α(OH)D3 remained relatively unchanged (Fig. 6-1B). Hence, it appeared that the

apparent EC50 estimated for 25(OH)D3 in absence of KTZ was only an apparent value due

to bioactivation and was not reflective of that for intact 25(OH)D3.

111

Figure 6-1. Vitamin D analogs are less potent than 1,25(OH)2D3 for VDR activity and are blunted by ketoconazole (KTZ) in the cell-based systems. (A) Luciferase activity (± SD) of HEK293 cells transfected with GAL4-hVDR-LBD in the presence of vehicle (0.1% ethanol) or various concentrations of 1,25(OH)2D3, vitamin D3, 25(OH)D3, 1α(OH)D3, and 1α(OH)D2 are shown. Transcriptional activities are as follows: 1,25(OH)2D3 >> 25(OH)D3 ≈ 1α(OH)D3 > 1α(OH)D2 >> vitamin D3. (B) In the presence of 10 µM KTZ, the luciferase activity (± SD) of HEK293 cells following 25(OH)D3 treatment was significantly blunted (EC50 increased from 200 to 608 nM), while activities following treatment with 1,25(OH)2D3 and 1α(OH)D3 remained unchanged. As such, the luciferase activity towards 25(OH)D3 treatment was attributed partially to its bioactivated species, 1,25(OH)2D3. Results are in relative light units (RLU). Similar results were obtained among three experiments and sampling was performed in triplicate.

112

6.4.2 Ketoconazole inhibition of 1α(OH)D3 bioactivation decreases its effects on VDR target genes in Caco-2 cells

Despite that the apparent EC50 values from the HEK293 luciferase reporter assay, in

absence of KTZ, were similar for 25(OH)D3 and 1α(OH)D3 in vitro (Fig. 6-1A), differences were observed with the Caco-2 cell system. Incubation of Caco-2 cells with 100 nM

1α(OH)D3 increased mRNA expression of VDR target genes: CYP24A1 (3200-fold), CYP3A4 (180-fold), TRPV6 (245-fold), and OATP1A2 (12-fold) when compared to those of 0.1% ethanol (control). The levels of induction were similar to those measured following

treatment with 100 nM 1,25(OH)2D3 (Fig. 6-2). Meanwhile, the gene expression remained relatively unchanged from vehicle control following treatment with 25(OH)D3. In presence

of 10 µM KTZ, 1α(OH)D3 induction was significantly decreased: the mRNA expression of VDR target genes CYP24A1 (73%), CYP3A4 (99%), TRPV6 (93%), and OATP1A2 (69%) fell in relation to those in absence of KTZ (Fig. 6-2). Induction levels of VDR target

genes for 1α(OH)D3 remained higher than those for 25(OH)D3, in absence or presence of

KTZ (Fig. 6-2). Addition of KTZ to cells treated with 1,25(OH)2D3 also reduced the mRNA expression of CYP3A4 (87%), TRPV6 (63%), and OATP1A2 (28%), however, CYP24A1

levels remained unchanged. The high EC50 of 25(OH)D3 that was estimated in the presence

of KTZ (Fig. 6-1B) suggests that this precursor is less potent than 1α(OH)D3, and its activity was significantly biased due to the presence of bioactivation enzymes in HEK293 cells.

113

Figure 6-2. Relative mRNA expression of VDR target genes in Caco-2 cells treated with 25(OH)D3, 1α(OH)D3, and 1,25(OH)2D3 ± ketoconazole (KTZ). Treatment with 100 nM 1α(OH)D3 significantly induced mRNA expression of VDR target genes to levels similar to those of 100 nM 1,25(OH)2D3: CYP24A1, CYP3A4, TRPV6, and OATP1A2. The induction of genes was significantly inhibited by addition of KTZ to levels similar to, but higher than that following 100 nM 25(OH)D3 treatment, except for CYP24A1 in the 1,25(OH)2D3-treated group. Data are mean ± SD of three experiments with sampling # performed in triplicate; denotes P < 0.05 for 1α(OH)D3- or 1,25(OH)2D3-treated vs. 0.1% ethanol (control) with 0 µM KTZ; * denotes P < 0.05 for 0 vs. 10 µM KTZ in respective treatment groups.

114

6.4.3 Basal expression of bioactivation enzymes in HEK293 vs. Caco- 2 cells

The discrepancies between 25(OH)D3 and 1α(OH)D3 behaviours in HEK293 vs. Caco-2 cells was suspected to be due to differences in bioactivation enzymes that may exist in each

cell system to activate these to 1,25(OH)2D3. For the bioactivation to 1,25(OH)2D3,

CYP27B1 is required for 25(OH)D3, while CYP2R1 and/or CYP27A1 is needed for

1α(OH)D3. In HEK293 cells where 25(OH)D3 and 1α(OH)D3 were shown to have similar VDR activities in vitro (Fig. 6-1A), the basal protein expression of CYP27B1 was significantly higher than that of Caco-2 cells (Fig. 6-3) where treatment with 25(OH)D3 was much less potent than 1α(OH)D3 (Fig. 6-2). Hence, the differences in observed potencies are indeed correlated to the higher expression of CYP27B1 in HEK293 cells vs. the extremely low expression of CYP27B1 in Caco-2 cells (Fig. 6-3). These findings were confirmed upon the addition of 10 µM KTZ to inhibit the CYP27B1-mediated activation

of 25(OH)D3, resulting in significantly reduced luciferase activity (Fig. 6-1B). The basal protein expression of CYP2R1 was higher in Caco-2 vs. HEK293 cells (Fig. 6-3); hence,

treatment with 1α(OH)D3 resulted in greater VDR target gene activation than 25(OH)D3 in the Caco-2 system.

Figure 6-3. Basal expression of CYP2R1 and CYP27B1 bioactivation enzymes in HEK293 vs. Caco-2 cells. HEK293 and Caco-2 cells were incubated with 0.1% ethanol (control) and CYP2R1 and CYP27B1 protein expression was measured. Caco-2 cells displayed higher basal expression of CYP2R1 whereas HEK293 cells showed a higher basal expression of CYP27B1. Data are mean ± SD of three experiments with sampling performed in triplicate.

115

6.4.4 Effects of vitamin D analogs on body weight and plasma ALT, bile acids, calcium, and phosphorus concentrations

After treatment with the vitamin D analogs, body weight and levels plasma ALT, bile acids, calcium, and phosphorus were compared to assess potential toxicity. The body weights of

mice treated with 1α(OH)D3 were decreased compared to those of WD controls (Table 6-

1), but those treated with vitamin D3, 25(OH)D3 and 1α(OH)D2 were unchanged.

Treatment with the relatively high dose of 25(OH)D3, but not with the other analogs, significantly increased ALT levels (1.7-fold), suggesting that the dose used was toxic when

compared to that of WD controls (Table 6-1). With the exception of vitamin D3, all groups showed non-significant increases in portal bile acid levels when compared to that of WD

controls (Table 6-1). Furthermore, vitamin D3 and 1α(OH)D2 treatment did not alter plasma

calcium nor phosphorus levels when compared to WD controls, while 25(OH)D3 and

1α(OH)D3 treatment resulted in elevation of plasma calcium (86 and 28%, respectively)

greater than or similar to levels previously reported following treatment with 1,25(OH)2D3

(39%). 1α(OH)D3 treatment also caused a decrease in plasma phosphorus (22%).

116

Table 6-1. Effects of vitamin D3, 25(OH)D3, 1α(OH)D3, and 1α(OH)D2 treatments on body weight and plasma ALT, bile acids, calcium, and phosphorus concentrations, compared with previously published 1,25(OH)2D3 data

Normal Diet Western Diet b Control Control Vitamin D3 25(OH)D3 1α(OH)D3 1α(OH)D2 1,25(OH)2D3 (corn oil) (corn oil) (1625 nmol/kg) (1248 nmol/kg) (1.75 nmol/kg) (1.21 nmol/kg) (6 nmol/kg) n=4 n=6 n=4 n=6 n=7 n=6 n=6 Body weight (g)c 27.2 ± 0.7a 28.8 ± 0.8 29.1 ± 1.0 28.1 ± 1.1 26.1 ± 0.4* 29.0 ± 0.6 24.0 ± 1.0* ALT (IU/ml) 15.1 ± 11.5 23.9 ± 6.7 31.9 ± 5.2 40.6 ± 18.1* 21.0 ± 5.8 17.8 ± 5.9 17.2 ± 2.7 Portal bile acids (µM) 43.9 ± 7.9 35.5 ± 18.3 28.6 ± 7.3 61.5 ± 35.9 49.3 ± 32.4 45.7 ± 21.2 31.0 ± 2.6 Calcium (mg/dl) 9.8 ± 0.5 9.2 ± 0.7 8.9 ± 0.6 17.1 ± 0.07* 11.8 ± 0.02* 10.8 ± 1.7 12.8 ± 0.5* Phosphorus (mg/dl) 15.9 ± 1.1 16.4 ± 1.8 16.4 ± 2.2 13.6 ± 2.8 12.8 ± 1.5* 18.1 ± 2.8 19.5 ± 0.6 a Data are mean ± SEM; samples were taken 50 h after last dose. b Data of Chow et al. (2014). c Body weight of mice before sacrifice. * P < 0.05 compared to Western diet control.

117

6.4.5 1,25(OH)2D3 concentrations increase in plasma and liver following treatment with 25(OH)D3 and 1α(OH)D3

In order to appraise the bioactivation of vitamin D analogs in mice in vivo and their

subsequent cholesterol lowering effects, levels of 1,25(OH)2D3 in plasma and liver were determined following treatment with the various vitamin D analogs. The basal plasma level of 1,25(OH)2D3 in control mice on a normal diet (127 ± 18 pM) was higher than mice fed the WD (74 ± 41 pM). Treatment with 1α(OH)D3 resulted in significantly higher plasma

concentrations of 1,25(OH)2D3 (162 ± 32 pM), while vitamin D3 and 25(OH)D3 failed to

alter plasma 1,25(OH)2D3 concentrations (Fig. 6-4A). In liver, basal 1,25(OH)2D3 levels were slightly lower in mice fed the normal diet (10.7 ± 1.9 pmol/kg tissue) compared with

the WD (17.3 ± 6.8 pmol/kg tissue). Treatment with 25(OH)D3 and 1α(OH)D3 increased

1,25(OH)2D3 levels in liver 3.4- and 1.8-fold, respectively (Fig. 6-4A). When 1,25(OH)2D3 levels following treatment were normalized to 1,25(OH)2D3 levels from the WD control, we found that treatment with all vitamin D analogs were able to increase plasma

1,25(OH)2D3 levels compared to mice administered directly with exogenous 1,25(OH)2D3

(Fig. 6-4B), a result of the intricate regulation of 1,25(OH)2D3 on itself (Chow et al., 2013b;

Quach et al., 2015). In liver, treatment with 25(OH)D3 and 1α(OH)D3 increased

1,25(OH)2D3 levels significantly compared to mice treated with 1,25(OH)2D3, while

vitamin D3 produced lower levels (Fig. 6-4B). Levels of 1,25(OH)2D2, the active

metabolite of 1α(OH)D2, could not be determined from this assay.

6.4.6 1α(OH)D3 treatment lowers plasma cholesterol in hypercholesterolemic mice

Plasma and liver cholesterol concentrations were significantly increased (1.4-fold and 17- fold, respectively) in control mice fed a WD compared to normal diet (Fig. 6-4C). When

normalized to WD control plasma cholesterol levels, mice treated with vitamin D3 and

1α(OH)D2 displayed significantly higher plasma cholesterol levels when compared with

1,25(OH)2D3-treated mice, while levels from 25(OH)D3 and 1α(OH)D3 treatment were similar to those in 1,25(OH)2D3 (Fig. 6-4D). Liver cholesterol levels normalized to WD

118

controls remained higher in mice treated with vitamin D3, 1α(OH)D3, and 1α(OH)D2 than in mice treated with 1,25(OH)2D3 (Fig. 6-4D). Meanwhile, treatment with 25(OH)D3 resulted in a similar decrease in liver cholesterol compared with 1,25(OH)2D3-treated mice (Fig. 6-4D). Data on the lack of cholesterol lowering effect in the liver upon treatment with

1α(OH)D2 contrasts the cholesterol lowering effect that was previously reported in the mouse kidney (Wang et al., 2011).

Figure 6-4. 1,25(OH)2D3 and cholesterol levels in plasma and liver in Western diet (WD)-fed mice treated with vitamin D analogs. Treatment with 1α(OH)D3 led to significantly higher 1,25(OH)2D3 concentrations in plasma and liver, while treatment with 25(OH)D3 increased only liver 1,25(OH)2D3 (A). When normalized to WD controls, all vitamin D analogs produced significantly higher plasma 1,25(OH)2D3 levels vs. treatment with exogenous 1,25(OH)2D3, while 25(OH)D3 and 1α(OH)D3 treatment further increased 1,25(OH)2D3 levels in liver (B). Treatment with 1α(OH)D3 significantly decreased plasma cholesterol, while treatment with 25(OH)D3 significantly decreased liver cholesterol. Additionally, plasma and liver cholesterol levels were significantly increased in WD-fed mice compared to normal diet controls (C). When normalized to WD controls, plasma cholesterol remained unchanged for mice treated with 1α(OH)D3 and 25(OH)D3 compared to 1,25(OH)2D3. In the liver, only mice treated with 25(OH)D3 displayed comparable cholesterol levels to those treated with 1,25(OH)2D3 (D). Data are mean ± SEM (n=3-7); †, P < 0.05 for normal diet vs. WD controls. *, P < 0.05 for WD control vs. treated mice. # , P < 0.05 for 1,25(OH)2D3-treated mice vs. vitamin D analog-treated mice. 1,25(OH)2D3 data and its respective WD control was taken from Chow et al. (2014) and unpublished data from our laboratory.

119

6.4.7 Plasma and liver cholesterol levels are inversely correlated with

1,25(OH)2D3

When levels of plasma cholesterol were plotted against plasma 1,25(OH)2D3, an inverse correlation, albeit shallow, was observed (Fig. 6-5A). Meanwhile, the inverse correlation was steeper and more significant when levels of liver cholesterol were plotted against liver

1,25(OH)2D3 (Fig. 6-5B), suggesting that bioactivation of vitamin D analogs to

1,25(OH)2D3 is of paramount importance at the tissue site where cholesterol synthesis occurs and where 1,25(OH)2D3-liganded VDR effects are exerted.

Figure 6-5. 1,25(OH)2D3 and cholesterol concentrations are inversely related. An inverse correlation was observed when plasma (A) and liver (B) cholesterol levels were correlated with their respective levels of 1,25(OH)2D3. Each datum point represents one mouse.

120

6.4.8 Effects of vitamin D analogs on hepatic nuclear receptors, transporters, and enzymes

Since treatment with 1,25(OH)2D3 was previously shown to lower cholesterol levels in hypercholesterolemic mice by down-regulation of Shp and induction of Cyp7a1 (Chow et al., 2014), we measured the expression of these cholesterol-regulating genes following treatment with various vitamin D analogs. Vdr and Fxr mRNA levels remained unchanged except for treatment with 1α(OH)D2, which significantly increased Fxr (Table 6-2). Shp mRNA levels were significantly increased with WD compared to normal diet controls,

while treatment with 25(OH)D3 and 1α(OH)D3 significantly decreased expression (45 and

50%, respectively) to levels similar to that following treatment with 1,25(OH)2D3 (Table

6-2). Correspondingly, treatment with 1α(OH)D3 significantly increased Cyp7a1 mRNA

(1.9-fold) and protein (1.6-fold) levels, while treatment with 25(OH)D3 increased protein expression (1.4-fold) when compared with WD controls. All other treatments failed to elicit any changes in Cyp7a1 (Table 6-2). The mRNA expression of cholesterol efflux transporters (Abca1, Abcg5, and Abcg8) was increased in WD controls when compared to normal diet controls, but remained relatively unchanged upon treatment with 25(OH)D3 and 1α(OH)D3. Overall, only treatment with 25(OH)D3 and 1α(OH)D3 displayed promising

cholesterol lowering effects, with greater therapeutic potential for 1α(OH)D3 since the dose

utilized for 1α(OH)D3 was 1/700-fold the dose of 25(OH)D3. Bioactivation of 25(OH)D3 and 1α(OH)D3 to 1,25(OH)2D3 in mice in vivo (Figs. 6-4, A and B) could explain the greater cholesterol lowering effects that were observed.

121

Table 6-2. Effects of vitamin D3, 25(OH)D3, 1α(OH)D3, and 1α(OH)D2 treatments on hepatic mRNA expression and Cyp7a1 protein expression, compared with previously published 1,25(OH)2D3 data

Normal Diet Western Diet b Control Control Vitamin D3 25(OH)D3 1α(OH)D3 1α(OH)D2 1,25(OH)2D3 (corn oil) (corn oil) (1625 nmol/kg) (1248 nmol/kg) (1.75 nmol/kg) (1.21 nmol/kg) (6 nmol/kg) n=4 n=6 n=4 n=6 n=7 n=6 n=6 Vdr 1.00 ± 0.33a 0.54 ± 0.13 0.28 ± 0.07 0.26 ± 0.06 0.35 ± 0.12 0.88 ± 0.13 0.58 ± 0.07 Fxr 1.00 ± 0.04 0.85 ± 0.08 0.93 ± 0.06 1.31 ± 0.22 0.91 ± 0.08 1.61 ± 0.09* 0.59 ± 0.07 Shp 1.00 ± 0.05 1.68 ± 0.12† 1.36 ± 0.19 0.94 ± 0.08* 0.86 ± 0.10* 1.44 ± 0.14 0.90 ± 0.19* Abca1 1.00 ± 0.13 1.47 ± 0.10† 1.65 ± 0.05 1.73 ± 0.12 1.46 ± 0.17 1.72 ± 0.32 1.05 ± 0.04 Abcg5 1.00 ± 0.18 2.50 ± 0.33† 4.52 ± 0.31* 3.86 ± 0.32 3.54 ± 0.28 5.21 ± 0.50* 0.91 ± 0.18 Abcg8 1.00 ± 0.23 3.29 ± 0.42† 5.87 ± 0.38* 2.51 ± 0.18* 3.16 ± 0.28 4.38 ± 0.76 0.66 ± 0.13 Cyp7a1 mRNA 1.00 ± 0.32 1.17 ± 0.18 1.23 ± 0.28 1.34 ± 0.35 1.97 ± 0.30* 1.10 ± 0.13 2.70 ± 0.98 Cyp7a1 protein 1.00 ± 0.15 0.95 ± 0.16 1.16 ± 0.23 1.34 ± 0.09* 1.54 ± 0.14* 0.90 ± 0.08 1.53 ± 0.14* a Data are normalized to normal diet control and represented as mean ± SEM; samples were taken 50 h after last dose. b Data from Chow et al. (2014). † P < 0.05 compared to normal diet control. * P < 0.05 compared to Western diet control.

122

6.5 Discussion

Recently, our laboratory identified that treatment with 1,25(OH)2D3 in mice in vivo could repress Shp for the induction of Cyp7a1 for cholesterol lowering (Chow et al., 2014). However, the

therapeutic potential of 1,25(OH)2D3 has been limited by the hypercalcemia associated with its use. While vitamin D analogs are logical alternatives, a number of studies have concluded that many of these analogs are less effective or elicit similar toxic effects when compared with

1,25(OH)2D3 (DeLuca et al., 2010; Chow et al., 2011b; Chow et al., 2013a). In the present study, we have provided compelling evidence to suggest that the potencies of vitamin D analogs in vitro

and in vivo are highly dependent upon their bioactivation to 1,25(OH)2D3. In vitro screening of

VDR activity following treatment with vitamin D analogs revealed that both 1α(OH)D3 and

25(OH)D3 were much more transcriptionally active when compared with vitamin D3 and

1α(OH)D2 and screening results in HEK293 cells further suggested that 1α(OH)D3 and 25(OH)D3

activities were similar (Fig. 6-1A). But the potency/activity of 25(OH)D3 became greatly diminished when KTZ was added for inhibition of bioactivation by 1α-hydroxylase (CYP27B1)

to 1,25(OH)2D3 (Fig. 6-1B). Furthermore, when Caco-2 cells were treated with KTZ to inhibit 25- hydroxylase (CYP2R1) activity, the induction of VDR target genes following treatment with

1α(OH)D3 was diminished to levels similar to those following treatment with 25(OH)D3 (Fig. 6- 2). Characterization of the basal expression of the bioactivation enzymes in both HEK293 and Caco-2 cell systems revealed that HEK293 cells displayed much higher levels of CYP27B1 while Caco-2 cells had higher levels of CYP2R1 (Fig. 6-3). Therefore, the in vitro potencies of vitamin D analogs could be incorrectly interpreted in cell systems that contain different levels of bioactivation enzymes.

In mice in vivo, we observed that 25(OH)D3 and 1α(OH)D3 were readily bioactivated to

1,25(OH)2D3 and demonstrated the most cholesterol lowering potential when compared with

vitamin D3 and 1α(OH)D2. 1α(OH)D3-mediated down-regulation of Shp mRNA expression and subsequent induction of Cyp7a1 mRNA and protein expression led to lower plasma cholesterol

levels. These changes were similar to those seen following 1,25(OH)2D3 treatment with a much

higher dose (Chow et al., 2014), suggesting that low doses of 1α(OH)D3 could be a potential non- toxic alternative. This observation could be explained by the continual bioactivation of 1α(OH)D3

123

to 1,25(OH)2D3, where the ubiquitous expression of CYP2R1 could sustain the 1,25(OH)2D3 pool in liver for a greater period of time following 1α(OH)D3 treatment. Moreover, there have been

several reports indicating that the liver is not the sole site for 25-hydroxylation of vitamin D3 (Olson et al., 1976), as substantial 25-hydroxylase activity has been previously demonstrated in kidney (Bergman and Postlind, 1990; Gascon-Barre et al., 2001), testes (Foresta et al., 2011), and intestine (Theodoropoulos et al., 2003). In addition to CYP2R1 and CYP27A1, CYP3A4 has also been shown to act as a 25-hydroxylase for 1α(OH)D3 in human liver microsomes (Gupta et al.,

2004). Since treatment with 1α(OH)D3 resulted in low blood phosphorus levels which has been previously reported to stimulate 1α-hydroxylase expression (Tanaka and DeLuca, 1984), this

induction may have also contributed to the rapid conversion of existing 25(OH)D3 levels to further

increase 1,25(OH)2D3 levels. While 1α-hydroxylase activity is thought to be exclusive to the kidney, it was previously suggested that extrarenal 1α-hydroxylase activity existed (Fleet et al.,

2008). Here, we found that treatment with a high dose of 25(OH)D3 resulted in a significant

increase in liver 1,25(OH)2D3 levels that resulted in the down-regulation of hepatic Shp mRNA and increase in Cyp7a1 protein expression to decrease liver cholesterol. However, treatment with

25(OH)D3 was less potent than 1α(OH)D3 in cholesterol lowering in vivo, even with the dose ratio 700-fold higher. Again, the in vivo data differ dramatically from those for the in vitro

transcriptional studies that predicted similar levels of activation between 25(OH)D3 and

1α(OH)D3. This discrepancy could be explained by the low levels of Cyp27b1 for the activation

of 25(OH)D3 in vivo.

Since the low dose 1α(OH)D3 appeared to display the greatest potential as a cholesterol lowering agent among the analogs examined, the potential for toxicity by monitoring body weights and levels of plasma ALT and calcium (Table 6-1) was further investigated. When vitamin D analogs

are administered, hypercalcemia is the major concern. Mice treated with 1α(OH)D3 vs.

1,25(OH)2D3 resulted in a lower decrease in body weight and less hypercalcemia. The calcium channels TRPV5/6 mediate calcium transport (Hoenderop et al., 1999) and are involved in apical renal and intestinal absorption of calcium, as evidenced by a significant reduction in calcium absorption and plasma calcium levels in Trpv6 knockout mice (Bianco et al., 2007; Cui et al., 2012). Activation of the VDR leads to induction of TRPV5 and TRPV6 via the genomic transcription of the VDREs and by the estrogen receptor in an attempt to increase plasma calcium

124

by reabsorption in the kidney and intestine (Hoenderop et al., 1999; Hoenderop et al., 2005). After

treatment with low dose 1α(OH)D3, plasma calcium concentrations were increased to a lesser

extent than previously seen with 1,25(OH)2D3 (1.28-fold vs. 1.42-fold, respectively) (Chow et al., 2014). In mice, induction of Trpv6 mRNA expression in the kidney was also significant following

25(OH)D3, 1α(OH)D3, and 1α(OH)D2 treatment. But the Trpv6 mRNA levels in the ileum

remained unchanged. In Caco-2 cells, treatment with 1α(OH)D3 induced TRPV6 mRNA levels and addition of KTZ to inhibit 25-hydroxylase activity significantly diminished the expression.

These results suggest that the conversion of 1α(OH)D3 to 1,25(OH)2D3 is mainly responsible for induction of TRPV6 and the associated hypercalcemia. Indeed, treatment with lithocholic acid

acetate (0.75 mmol/kg i.p. q2d x 4), an alternate VDR ligand that does not convert to 1,25(OH)2D3, did not trigger hypercalcemia (Ishizawa et al., 2008) while still increasing Cyp7a1 protein expression (2-fold) and modestly decreasing plasma cholesterol levels by 14% in mice (Fig. A3-

2; Appendix III). Thus, targeting the VDR with compounds that do not form 1,25(OH)2D3 may present opportunities as novel therapies for patients with hypercholesterolemia. Alternatively, the pharmacokinetic/pharmacodynamic models for the optimization of doses, dosing regimens, and/or

alternate routes of administration of 1,25(OH)2D3 (Ramakrishnan et al., 2016) or other vitamin D analogs could also be used to help predict cholesterol lowering efficacy while avoiding hypercalcemia.

6.6 Acknowledgments

This work was supported by the Canadian Institutes of Health Research (KSP), the National Sciences and Engineering Research Council of Canada (HPQ), and the Ontario Graduate Scholarship Program (HPQ). We thank Dr. Edwin Chow for his assistance in animal handling and providing his expertise in several laboratory techniques.

125

6.7 Supplementary material

Supplementary Table S6-1. Mouse and human primer sequences for qPCR

Gene Bank Forward Sequence (5'→3') Reverse Sequence (5'→3') Number mAbca1 NM_013454.3 CGTTTCCGGGAAGTGTCCTA CTAGAGATGACAAGGAGGATGGA mAbcg5 NM_031884.1 TCAATGAGTTTTACGGCCTGAA GCACATCGGGTGATTTAGCA mAbcg8 NM_026180.2 TGCCCACCTTCCACATGTC ATGAAGCCGGCAGTAAGGTAGA mAsbt NM_011388 GATAGATGGCGACATGGACCTC CAATCGTTCCCGAGTCAACC mCyp24a1 NM_009996 CTGCCCCATTGACAAAAGGC CTCACCGTCGGTCATCAGC mCyp7a1 NM_007824 AGCAACTAAACAACCTGCCAGTACTA GTCCGGATATTCAAGGATGCA mCyclophilin X58990 GGAGATGGCACAGGAGGAA GCCCGTAGTGCTTCAGCTT mFgf15 NM_008003 ACGGGCTGATTCGCTACTC TGTAGCCTAAACAGTCCATTTCCT mFxr NM_009108 CGGAACAGAAACCTTGTTTCG TTGCCACATAAATATTCATTGAGATT mMdr1a NM_011076 TACGACCCCATGGCTGGATC GGTAGCGAGTCGATGAACTG mMrp4 NM_001163675.1 GGTTGGAATTGTGGGCAGAA TCGTCCGTGTGCTCATTGAA mShp NM_011850 CAGCGCTGCCTGGAGTCT AGGATCGTGCCCTTCAGGTA mTrpv6 NM_022413 ATCGATGGCCCTGCCAACT CAGAGTAGAGGCCATCTTGTTGCTG mVdr NM_009504 GAGGTGTCTGAAGCCTGGAG ACCTGCTTTCCTGGGTAGGT mVillin NM_009509 TCCTGGCTATCCACAAGACC CTCTCGTTGCCTTGAACCTC hCYP24A1 NM_000782.4 CAGCGAACTGAACAAATGGTCG TCTCTTCTCATACAACACGAGGCAG hCYP27B1 NM_000785.3 CAGACAAAGACATTCATGTGGG GTTGATGCTCCTTTCAGGTAC hCYP2R1 NM_024514.4 CAGCCTCATCCGAGCTTC CCACAGTTGATATGCCTCCA hCYP3A4 NM_017460.5 CATTCCTCATCCCAATTCTTGAAGT CCACTCGGTGCTTTTGTGTATCT hGAPDH NM_002046.4 GAAGGTGAAGGTCGGAGTC GAAGATGGTGATGGGATTTC hOATP1A2 NM_021094.3 TGGGGAACTTTGAAATGTGG AAGGCTGGAACAAAGCTTGA hTRPV6 NM_018646.4 GGTTCCTGCGGGTGGAA CCTGTGCGTAGCGTTGGAT m, mouse; h, human a Detects both mMdr1a and mMdr1b

6.7.1 Effects of vitamin D analogs on ileal and renal nuclear receptors, transporters, and enzymes

Other VDR-related effects in the ileum and kidney, tissues with high VDR distribution, were further examined. Control mice fed a WD displayed significantly increased Shp, Fgf15, Abca1, Abcg5, and Abcg8 mRNA levels and decreased Fxr and Cyp24a1 mRNA levels when compared with normal diet controls, while Vdr, Asbt, and Trpv6 mRNA expression remained unchanged between the control groups (Table S6-2). As expected for VDR activation, 25(OH)D3 and

1α(OH)D3 treatment resulted in significantly higher Cyp24a1 mRNA expression, the signature

VDR target gene. Meanwhile, treatment with 25(OH)D3, 1α(OH)D3, and 1α(OH)D2 resulted in

126

significantly decreased mRNA levels of Shp and Fgf15, negative regulators of Cyp7a1. While

previous studies have shown 1,25(OH)2D3-mediated up-regulation of Asbt in the rat intestine (Chen et al., 2006; Chow et al., 2009), no significant change in Asbt mRNA expression was observed in the mouse, suggesting that the effects due to the Fxr-Shp-Lrh-1 cascade for inhibition of Asbt must have counterbalanced that of the Vdr for induction of Asbt. Ileal Trpv6 mRNA

expression was also unchanged with 25(OH)D3 and 1α(OH)D3, likely due to the measuring of mRNA levels 50 h post-injection which was past the peak time of induction (Chow et al., 2013b),

whereas ileal Trpv6 was induced upon treatment with 1α(OH)D2. The mRNA expression of cholesterol efflux transporters (Abca1, Abcg5, and Abcg8) was significantly decreased following

treatment with 25(OH)D3 and 1α(OH)D3 when compared to WD controls.

In the kidney, a VDR-rich organ, mRNA expression of Vdr and target genes including Cyp24a1, Mdr1, Mrp4, and Trpv6 were unchanged with the WD (Table S6-3). Concomitant to the treatment with the vitamin D analogs in mice fed the WD, there were significant increases in Vdr and

Cyp24a1 mRNA expression levels when compared to WD controls. Vitamin D3 did not elicit

changes in Mdr1, Mrp4, or Trpv6 mRNA levels, while a similar dose of 25(OH)D3 resulted in significant induction of these genes. The effects of 1α(OH)D2 and 1α(OH)D3, were significant,

with effects of 1α(OH)D2 > 1α(OH)D3.

127

Supplementary Table S6-2. Effects of vitamin D3, 25(OH)D3, 1α(OH)D3, and 1α(OH)D2 treatments on ileal mRNA expression, compared with previously published 1,25(OH)2D3 data

Normal Diet Western Diet Control Control Vitamin D 25(OH)D 1α(OH)D 1α(OH)D 1,25(OH) D b 3 3 3 2 2 3 (corn oil) (corn oil) (1625 nmol/kg) (1248 nmol/kg) (1.75 nmol/kg) (1.21 nmol/kg) (6 nmol/kg) n=4 n=6 n=4 n=6 n=7 n=6 n=6 Vdr 1.00 ± 0.11a 1.17 ± 0.04 1.13 ± 0.09 1.17 ± 0.10 0.80 ± 0.01 0.89 ± 0.04 0.74 ± 0.05 Fxr 1.00 ± 0.07 0.59 ± 0.05† 0.74 ± 0.10 1.13 ± 0.19* 0.36 ± 0.03* 0.54 ± 0.09 0.65 ± 0.05 Shp 1.00 ± 0.25 13.8 ± 0.93† 16.1 ± 3.25 3.50 ± 1.44* 4.66 ± 0.50* 4.12 ± 0.83* 8.54 ± 2.36 Fgf15 1.00 ± 0.23 4.79 ± 0.43† 4.84 ± 0.75 1.30 ± 0.55* 3.19 ± 0.30* 2.66 ± 0.50* 3.63 ± 0.71 Cyp24a1 1.00 ± 0.11 0.40 ± 0.07† 0.66 ± 0.13 11.4 ± 2.96* 0.73 ± 0.07* 1.32 ± 0.44* 8.37 ± 1.65* Trpv6 1.00 ± 0.09 1.18 ± 0.34 1.09 ± 0.08 1.48 ± 0.42 0.96 ± 0.15 3.27 ± 0.13* 1.84 ± 0.11 Abca1 1.00 ± 0.15 3.82 ± 0.36† 5.39 ± 0.89 1.81 ± 0.36* 2.43 ± 0.23* 3.86 ± 0.56 2.09 ± 0.26* Abcg5 1.00 ± 0.05 2.30 ± 0.13† 2.43 ± 0.31 1.11 ± 0.11* 1.56 ± 0.05* 2.69 ± 0.40 1.29 ± 0.16* Abcg8 1.00 ± 0.03 2.14 ± 0.23† 2.23 ± 0.34 1.09 ± 0.15* 1.23 ± 0.05* 2.08 ± 0.44 1.20 ± 0.13 a Data are normalized to normal diet control and represented as mean ± SEM; samples were taken 50 h after last dose. b Data from Chow et al. (2014). † P < 0.05 compared to normal diet control. * P < 0.05 compared to Western diet control.

128

Supplementary Table S6-3. Effects of vitamin D3, 25(OH)D3, 1α(OH)D3, and 1α(OH)D2 treatments on renal mRNA expression, compared with unpublished 1,25(OH)2D3 data

Normal Diet Western Diet b Control Control Vitamin D3 25(OH)D3 1α(OH)D3 1α(OH)D2 1,25(OH)2D3 (corn oil) (corn oil) (1625 nmol/kg) (1248 nmol/kg) (1.75 nmol/kg) (1.21 nmol/kg) (6 nmol/kg) n=4 n=6 n=4 n=6 n=7 n=6 n=6 Vdr 1.00 ± 0.05a 0.90 ± 0.07 1.37 ± 0.18* 1.58 ± 0.19* 1.47 ± 0.13* 2.05 ± 0.27* 2.25 ± 0.20* Cyp24a1 1.00 ± 0.18 0.84 ± 0.16 4.27 ± 0.85* 7.69 ± 0.29* 9.58 ± 0.25* 15.93 ± 1.26* 12.76 ± 0.62* Mdr1a 1.00 ± 0.22 0.89 ± 0.09 1.34 ± 0.15 8.56 ± 0.75* 4.78 ± 0.43* 5.28 ± 0.57* 3.78 ± 0.27* Mrp4 1.00 ± 0.09 0.79 ± 0.06 0.93 ± 0.17 1.61 ± 0.09* 1.14 ± 0.08* 1.61 ± 0.21* 1.31 ± 0.08* Trpv6 1.00 ± 0.09 1.06 ± 0.09 0.96 ± 0.06 5.29 ± 0.48* 2.23 ± 0.19* 2.58 ± 0.41* 2.48 ± 0.30* a Data are normalized to normal diet control and represented as mean ± SEM; samples were taken 50 h after last dose. b unpublished data from our laboratory. † P < 0.05 compared to normal diet control. * P < 0.05 compared to Western diet control.

129

6.8 Statement of significance of chapter 6

The therapeutic utility of 1,25(OH)2D3 is limited by its propensity to induce hypercalcemia. The development of vitamin D analogs that elicit desirable therapeutic effects without causing toxicity

has been a focus of research for many years. The immediate precursors of 1,25(OH)2D3, vitamin

D3 and 25(OH)D3, as well as the prodrug 1α(OH)D3. Although these analogs may bind to the VDR

directly, it is at much lower affinity when compared with 1,25(OH)2D3. Upon administration, these

compounds are bioactivated to 1,25(OH)2D3, the extent of which is dependent upon the activity of

1α-hydroxylase [CYP27B1; for conversion of 25(OH)D3] and 25-hydroxylase [CYP2R1; for

conversion of 1α(OH)D3]. Hence, the potential therapeutic effects (i.e. induction of Cyp7a1 and lowering cholesterol) that are observed following administration of these analogs could be a combined result of direct binding to the VDR as well as bioactivation to 1,25(OH)2D3.

Although the data on reporter assays in HEK293 cells suggested that 25(OH)D3 and 1α(OH)D3 would have similar potency, our in vivo mouse study showed that 1α(OH)D3 was much more

potent than 25(OH)D3, even at a 700-fold lower dose, at inducing VDR target genes. We found that this discrepancy between in vitro and in vivo results stems from a difference in expression of bioactivation enzymes among the systems. In Caco-2 cells, we demonstrated that the addition of

KTZ, a CYP inhibitor, suppressed 1α(OH)D3-mediated activation of VDR target genes, confirming that the VDR potency of the analogs is highly dependent upon bioactivation to

1,25(OH)2D3. Meanwhile, we observed that treatment of Caco-2 cells with 25(OH)D3 resulted in less pronounced effects on VDR target genes, and we verified this with the presence of low CYP27B1 levels in the cell line. HEK293 cells, found to have significantly higher CYP27B1

expression when compared with Caco-2 cells, explained our initial observation that 25(OH)D3 and

1α(OH)D3 would have similar potency. These observations were confirmed by the addition of

KTZ to the reporter assays in HEK293 cells, which greatly diminished the activity of 25(OH)D3. Therefore, discrepancy in the levels of bioactivation enzymes in the different cell systems could explain the data. In vivo, direct measurement of 1,25(OH)2D3 levels following administration of

analogs showed that 1α(OH)D3 was producing greater levels of 1,25(OH)2D3 in plasma and liver,

while 25(OH)D3 was producing 1,25(OH)2D3 in liver only. This finding could explain the

enhanced efficacy of 1α(OH)D3 over 25(OH)D3 for eliciting cholesterol lowering effects, even

130

when given at a 700-fold lower dose. Hence, we recommend careful consideration of the basal expression of these bioactivation enzymes in cell-based systems prior to the interpretation of results following treatment with vitamin D analogs.

My contributions to this chapter include conducting experiments for all of the in vivo work and writing the paper. Paola Bukuroshi was responsible for design of in vitro experiments, treatment and harvest of cells, determination of mRNA and protein for the in vitro study, and writing the paper. Tamara Dzekic contributed to the in vivo study. Lilia Magomedova performed the luciferase reporter assay. Carolyn L. Cummins contributed to design of in vitro experiments. K. Sandy Pang contributed to design of experiments, analyses of data, and writing the paper.

131

Chapter 7

Pharmacokinetic/Pharmacodynamic Modeling to Describe

1α,25-Dihydroxyvitamin D3-Mediated Cholesterol Lowering

Contributions of H. P. Quach excerpted from:

PKPD Modeling to Predict Altered Disposition of 1α,25-Dihydroxyvitamin D3 in Mice Due to Dose-Dependent Regulation of CYP27B1 on Synthesis and CYP24A1 on Degradation

Holly P. Quach1, Qi Joy Yang1, Edwin C.Y. Chow1, Donald E. Mager2, Stacie Y. Hoi1, and K. Sandy Pang1

and

Physiologically-Based Pharmacokinetic-Pharmacodynamic Modeling of

1α,25-Dihydroxyvitamin D3 in Mice

Vidya Ramakrishnan2, Qi Joy Yang1, Holly P. Quach1, Yanguang Cao2, Edwin C.Y. Chow1, Donald E. Mager2, and K. Sandy Pang1

1Department of Pharmaceutical Sciences, Leslie Dan Faculty of Pharmacy, University of Toronto, Toronto, Ontario, Canada

2Department of Pharmaceutical Sciences, University at Buffalo, State University of New York, Buffalo, New York, USA

Br J Pharmacol 2015; 172:3611-3626 Drug Metab Dispos 2016; 44:189-208

132

7.1 Abstract

Concentrations of 1α,25-dihydroxyvitamin D3 [1,25(OH)2D3], the active ligand of the vitamin D receptor, are tightly regulated by Cyp27b1 for synthesis in the kidney and Cyp24a1 for degradation. However, the dose-dependent pharmacokinetic (PK)-pharmacodynamic (PD)

relationship between these enzymes and 1,25(OH)2D3 concentrations has not been characterized.

The nonlinearity in the PK of 1,25(OH)2D3 was evaluated after administration of single (2, 60 and 120 pmol) and repeated (2 and 120 pmol q2d x3) i.v. doses to C57BL/6 mice. In parallel, mRNA expression of Cyp27b1 and Cyp24a1 was examined by quantitative PCR and 1,25(OH)2D3 concentrations were determined by enzyme-immunoassay. Fitting with a simple two-compartment model revealed decreasing net synthesis rates and increasing total clearances with dose, consistent with a dose-dependent downregulation of renal Cyp27b1 and the induction of renal/intestinal Cyp24a1 mRNA expression. Further incorporation of PD parameters for inhibition of Cyp27b1 and induction of Cyp24a1 to the simple two-compartment model greatly improved the fit to all doses. Additionally, an indirect response model, which considered the synthesis and degradation of enzymes, was consistent with the data in describing the PK and PD profiles. Since full physiologically-based PK (PBPK) modeling better describes the metabolism and feedback control occurring in multiple organs, PBPK models that incorporated a traditional intestine model [PBPK(TM)-PD] or a segregated flow model [PBPK(SFM)-PD] were used to predict the i.p. data of Chow et al. (2013b) and the i.v. data. Overall, the PBPK(SFM)-PD model was more consistent with both sets of data when compared with the PBPK(TM)-PD model. The structure of the PBPK(SFM)-PD model was extended to incorporate the cholesterol lowering effect that was observed upon repeated i.p. administration of 1,25(OH)2D3.

7.2 Introduction

Vitamin D, formed from 7-dehydrocholesterol in skin upon exposure to sunlight, is metabolized

by CYP2R1 and CYP27A1 in liver to its major circulating form, 25-hydroxyvitamin D3

[25(OH)D3]. This relatively inactive metabolite is transported by the vitamin D binding protein (DBP) for activation by 1α-hydroxylase (CYP27B1) in kidney to form the active ligand of the

133

vitamin D receptor (VDR), 1α,25-dihydroxyvitamin D3 [1,25(OH)2D3] (Jones et al., 1998). A

major physiological role of 1,25(OH)2D3 is to regulate plasma calcium concentrations through the calcium ion channels, TRPV5 and TRPV6, in the kidney and intestine (den Dekker et al., 2003) and the calcium-sensing receptor (Carrillo-Lopez et al., 2008). Continuous bone turnover, including resorption of existing bone and deposition of new bone, is another process that is

stimulated by 1,25(OH)2D3 and the parathyroid hormone (PTH) (Jones et al., 1998; Hoenderop et al., 2005).

Calcium and 1,25(OH)2D3 homeostasis is tightly controlled by 1,25(OH)2D3, calcium and PTH

(Shinki et al., 1992; Masuda et al., 2005; Turunen et al., 2007). Plasma 1,25(OH)2D3 concentrations are regulated by two major enzymes: CYP27B1 for synthesis and CYP24A1 for degradation. CYP27B1, expressed predominantly in kidney, responds positively to PTH at low plasma calcium concentrations (Shinki et al., 1992), but is down-regulated by high concentrations of 1,25(OH)2D3 (Brenza and DeLuca, 2000; Turunen et al., 2007). CYP24A1, distributed

abundantly in the kidney and intestine, is responsible for the metabolism of 25(OH)D3 to 24,25- dihydroxyvitamin D3 and 1,25(OH)2D3 to 1α,24,25-trihydroxyvitamin D3 (Holick et al., 1972;

Kumar et al., 1978; Halloran and Castro, 1989). Because elevated concentrations of 1,25(OH)2D3 are known to cause hypercalcemia (Jones et al., 1987; Makin et al., 1989), CYP24A1 expression in the kidney is up-regulated as a feedback mechanism to increase 1,25(OH)2D3 catabolism and

reduce 1,25(OH)2D3 and 25(OH)D3 stores (Clements et al., 1992). In contrast, intestinal CYP24A1

is regulated by 1,25(OH)2D3 and not PTH (Henry, 2001), suggesting that induction of intestinal CYP24A1 is an acute response to the VDR (Akeno et al., 1994).

Pharmacokinetic (PK) studies on 1,25(OH)2D3 are challenging due to assay sensitivity in

measuring low 1,25(OH)2D3 concentrations and studies in rodents are further hampered by the limited plasma volume for sampling. Masuda et al. (2005) examined the decay of radiolabelled

1,25(OH)2D3 over 96 h in Cyp24a1(+/-) and Cyp24a1(-/-) mice, confirming that Cyp24a1 is the

major enzyme involved in the metabolism of 1,25(OH)2D3. Cyp24a1(-/-) mice exhibited a longer

t1/2 compared with Cyp24a1(+/-) mice. In a phase I clinical trial, where 2 to 10 µg 1,25(OH)2D3 was administered s.c., the derived t1/2 proved to be ill-defined due to inadequate sampling over 12 h (Smith et al., 1999). In a human study in which prolonged sampling was conducted following

p.o. and i.v. doses of 4 µg 1,25(OH)2D3, a t1/2 of 26 h was observed, along with a plasma clearance 134

(dose/AUC∞) of 0.17 mL·min-1·kg-1 and bioavailability of 0.71 after 72 h of sampling (Brandi et al., 2002). C3H/HeJ mice treated with 0.125 or 0.5 µg 1,25(OH)2D3 i.p., with sampling up to 24

h, produced an apparent clearance (dose/AUC0→24), but a debatable terminal t1/2 due to limited

sampling (Muindi et al., 2004). Chow et al. (2013b) reported an apparent terminal t1/2 of 6.8 h after sampling for 48 h in mice treated with 0.05 µg 1,25(OH)2D3 i.p. and showed that both plasma and

tissue 1,25(OH)2D3 concentrations fell below basal levels at 24 h due to the induction of Cyp24a1. None of these studies provided an in-depth interpretation of the PK of exogenously administered

1,25(OH)2D3 by including the net rate of synthesis (Rsyn) of endogenous 1,25(OH)2D3 nor accounted for pharmacodynamic (PD) changes on the inhibition of Cyp27b1 or induction of Cyp24a1.

In the i.v. study, we examined the PK and PD changes driven by the exogenous administration of

1,25(OH)2D3 in relation to basal concentrations of 1,25(OH)2D3. Single and repeated i.v. doses

were administered to mice to appraise the dose- and time-dependent PK of 1,25(OH)2D3. Fitting with a simple two-compartment model yielded a decreasing Rsyn and increasing total plasma

clearance (CLtotal), observations consistent with the inhibition of Cyp27b1 and induction of Cyp24a1 with dose. Inclusion of parameters associated with PD changes in mRNA expression for the downregulation of Cyp27b1 and the induction of Cyp24a1, obtained upon regression of mRNA expression fold change (FC) of enzymes in tissue versus 1,25(OH)2D3 plasma concentration, significantly improved model fitting criteria. Then, an indirect response model, with sigmoidal

Emax/Imax and EC50/IC50, was able to provide similar parameters as the PKPD model, and predicted the temporal PK and PD data for the different doses administered. The composite data show that

changes in the PD of Cyp27b1 and Cyp24a1 with increasing 1,25(OH)2D3 doses resulted in altered

PK of 1,25(OH)2D3.

Since physiologically-based PK (PBPK) modeling better describes metabolism and feedback control occurring in multiple organs, PBPK models that incorporated a traditional intestine model [PBPK(TM)-PD] or a segregated flow model [PBPK(SFM)-PD] with a lower flow perfusing the enterocyte region, and other tissues including the kidney, liver, and brain, were utilized to fit the i.p. data of Chow et al. (2013b), which contained rich data sets on changes in enzyme expression. The two models described varying perfusion patterns to the intestine: the TM, in which the entire blood flow perfuses the intestine tissue as a whole, and the SFM, which describes the intestine as 135

an active enterocyte region perfused by a low and partial intestinal blood flow (5-30%), and an inactive serosal region, perfused by the remaining flow (Cong et al., 2000). The SFM describes a smaller extent of intestinal elimination with i.v. compared with oral dosing due to shunting of flow to the enterocytes, delimiting access of the bulk of the drug after i.v. dosing (Cong et al., 2000), and predicts that metabolism by the intestine may be route-dependent. The published fit showed that the PBPK(SFM)-PD model was indeed superior in describing the changes in the PD of

Cyp27b1 and Cyp24a1 that altered the PK of 1,25(OH)2D3. Hence, we presently utilized the developed PBPK(SFM)-PD model (Appendix IV; Ramakrishnan et al., 2016) to describe the cholesterol lowering effect that was previously reported by Chow et al. (2014). In this study, it was shown that mice fed the high fat/high cholesterol diet were hypercholesterolemic, an effect

that was ameliorated upon treatment with 1,25(OH)2D3 that induced cholesterol 7α-hydroxylase (Cyp7a1) via suppression of the small heterodimer partner (Shp) (Chapter 4; Chow et al., 2014).

7.3 Materials and methods

7.3.1 Materials

1,25(OH)2D3 in powder form was obtained from Sigma-Aldrich (Mississauga, ON). The enzyme- immunoassay (EIA) kit (Cat# AC-62F1) for 1,25(OH)2D3 measurements was manufactured by Immunodiagnostics Systems Inc. (Scottsdale, AZ) and purchased from Inter Medico (Markham, ON). All other reagents were obtained from Sigma-Aldrich and Fisher Scientific (Mississauga, ON).

7.3.2 In vivo pharmacokinetic study for i.v. dosing of 1,25(OH)2D3

The concentration of 1,25(OH)2D3 in anhydrous ethanol was assayed spectrophotometrically at 265 nm (UV-1700, Shimadzu Scientific Instruments, Mandel Scientific, Guelph, ON) and diluted with sterile 0.9% saline containing 1% ethanol. Male C57BL/6 mice (8-weeks old), weighing 25.5 ± 1.6 g (mean ± SD), were purchased from Charles River Canada (Saint-Constant, QC). Mice were given water and food ad libitum and maintained under a 12:12-h light and dark cycle in accordance with approved protocols by the Animal Care and Use Committee at the University of Toronto. All

136

studies involving animals were conducted in accordance with the ARRIVE guidelines for reporting experiments involving animals (Kilkenny et al., 2010). A total of 130 mice were used in the experiments described to provide the kinetic and enzyme data for model building. Mice were randomly assigned to treatment with single (0, 2, 60 or 120 pmol) or repeated (0, 2 or 120 pmol q2d x3) i.v. doses of 1,25(OH)2D3 on Days 0, 2 and 4 at 0900 h. Serial blood sampling from the saphenous vein was performed at 1, 5, 15, 30 or 60 min. Thereafter, mice were anesthetized with ketamine and xylazine i.p. (150 and 10 mg·kg-1, respectively) before blood collection by cardiac puncture with a 1 mL syringe-23G 3/4” needle set that was pre-rinsed with heparin (1000 IU·mL- 1). The depth of anesthesia was assessed by monitoring the heart rate and pedal reflex. Tissues were harvested at each sampling point (3, 6, 9, 12, 24 and 48 h) from the treated mice (n = 3-4 per time point). For the vehicle-treated group (n = 9), sampling was conducted at 0 h on Days 0 and 4 and averaged, as described previously (Chow et al., 2013b), to provide basal concentrations. Plasma was obtained by centrifugation of blood at 3000 g for 10 min. After flushing the lower vena cava with ice-cold saline, the kidneys and ileum (6 cm proximal to the ileocecal junction) were removed over ice, as outlined previously (Chow et al., 2011a). Samples were snap-frozen in liquid nitrogen and stored at -80°C.

7.3.2.1 Plasma 1,25(OH)2D3 analysis

Plasma 1,25(OH)2D3 concentrations were measured by EIA (Chow et al., 2013b).

7.3.2.2 Quantitative real-time PCR

Total RNA, obtained from kidney tissues and scraped ileal enterocytes, was extracted using the TRIzol extraction method (Sigma-Aldrich) in accordance with the manufacturer’s protocol, with modifications (Chow et al., 2011a). A total of 1.5 µg of cDNA was synthesized from RNA using the high capacity cDNA reverse transcription kit (Applied Biosystems® by Life Technologies, Burlington, ON) and qPCR was performed with SYBR Green detection system. For the i.v. study, kidney and intestinal mRNA data were normalized to Cyclophilin and Villin, respectively, for calculation of the relative change in gene expression (Chow et al., 2009).

137

7.3.2.3 Analysis of i.v. data by PKPD modeling

7.3.2.3.1 Non-compartmental analysis

The AUC from 0 to 48 h (AUC0→48h) was estimated by the trapezoidal rule. The extrapolated area

from the last datum point to time infinity (AUC48h→∞) was calculated upon dividing the measured plasma concentration, C48h, by the terminal slope (β). Other parameters included: t1/2β or terminal

half-life, estimated from data between 6 and 48 h and calculated as 0.693/β, and CLtotal, estimated

as dose/AUC∞.

7.3.2.3.2 Estimation of PD parameters for inhibition and induction

The dynamic changes of the mRNA expression for inhibition or induction of VDR target genes were given by the FC in gene expression after treatment with 1,25(OH)2D3, normalized to the basal levels (vehicle-treated mice).

For Cyp27b1, the inhibition function or FC of Cyp27b1 (Cyp27b1FC) was: IC Cyp27b1 =(1-max p ) FC (1) IC50 +C p

For Cyp24a1, the induction function or FC of Cyp24a1 (Cyp24a1FC) was: EC Cyp24a1 =(1+max p ) FC (2) EC50 +C p

where Cp is the plasma 1,25(OH)2D3 concentration, Imax and Emax are the maximal FC for inhibition

and induction factors, respectively, and IC50 and EC50 are the plasma concentrations that result in

50% of Imax and Emax (Mager et al., 2009). Upon presentation of the FC vs. the assayed

1,25(OH)2D3 concentrations, estimates of Emax, EC50, Imax and IC50 were obtained by non-linear regression of either data from single or repeated doses with Equations 1 and 2 using Scientist® (version 2.0; Micromath, St. Louis, MO). The mRNA data are expressed as mean ± SEM. One- way ANOVA and a post hoc Tukey honest significant difference test were used to evaluate differences between mean mRNA expression of groups at each time point using GraphPad Prism Software (version 6; GraphPad Software Inc., La Jolla, CA).

138

7.3.2.3.3 Fitting of i.v. data with compartmental modeling

Fitting was conducted with Scientist with appropriate weighting schemes (unity, 1/observation and 1/observation2) or with ADAPT5 (version 5; Biomedical Simulations Resource, University of Southern California, Los Angeles, CA). Various compartmental models were compared (Fig. 7- 1), including: (1) simple two-compartment model, (2) PKPD model, and (3) the indirect response model. The goodness of fit was appraised by the weighted sum of square residuals (WSSR), Akaike information criterion (AIC), and SD of parameter estimate, while the F-test was used for comparing the models. Significance was defined as P < 0.05.

Figure 7-1. Fitting of i.v. 1,25(OH)2D3 data using (A) a simple two-compartment model, (B) PKPD model, and (C) an indirect response model. For the compartmental models, it is assumed that synthesis and elimination of 1,25(OH)2D3 are occurring from the central compartment.

139

Assuming stationary kinetics and dynamics, we estimated the volume of the central compartment

(V1), net synthesis rate (Rsyn) controlled by Cyp27b1, and the micro-rate constants (k12, k21, k10) during the first 48 h of sampling for the single dose data. The fit for each of the single 2, 60 or 120 pmol dose levels was obtained with a simple two-compartment model (Fig. 7-1A and equations in

Appendix A). The steady-state volume of distribution (Vss) was estimated as V1·(1+k12/k21).

Values of k12 for the 60 and 120 pmol doses are expressed as multiples of k12, relative to the 2

pmol dose, for the combined fit of all data (1.5x and 5.7x the k12 value of the 2 pmol dose). Parameter estimates obtained from the single doses were used as initial estimates for the combined fit to all data upon repeated dosing.

Because nonlinearity was observed among the escalating doses, the underlying reason for this nonlinearity was examined. Parameters governing changes in Cyp27b1 and Cyp24a1 expression (Equations 1 and 2) were incorporated into the two-compartment model (Fig. 7-1B and Appendix

B) to modify the synthesis pathway (Rsyn) and elimination (rate constant, k10), respectively. Based on the assumption that there was no downregulation of Cyp27b1 nor induction of Cyp24a1 for the

2 pmol dose, fitted parameters for this dose (V1, k12, k21, Rsyn and k10) from the two-compartment

model were used as initial estimates, together with averaged estimates of Imax, IC50, Emax and EC50 (Table 7-3), for combined fitting of data for the first doses (2, 60 and 120 pmol) and data from all doses. Fitting was repeated upon addition of scaling factors, then Hill coefficients (Appendix B).

Again, k12 values for the 60 and 120 pmol doses were scaled to account for possible changes in

the distribution of 1,25(OH)2D3 in the simultaneous fits. This PKPD model was compared with the simple two-compartment model using the F-test (Boxenbaum et al., 1974).

The indirect response model (Fig. 7-1C), which describes an indirect mechanism of action and

incorporates transit compartments (Atransit1 and Atransit2) containing a time delay function (τ) (Appendix C), relates temporal differences between drug concentrations and responses (Mager et al., 2003). The indirect response model was used to explain the time delay for the Cyp27b1 effects, assuming Cyp27b1 is indirectly stimulated by PTH-induced activation of the VDR. Similar to the PKPD model, scaling factors and Hill coefficients were incorporated for fitting purposes.

140

7.3.3 The PBPK(TM)-PD and PBPK(SFM)-PD models

7.3.3.1 PBPK-PD modeling

PBPK-PD models, consisting of 11 compartments, with 6 representing various tissues [plasma, brain, liver, kidney, ileum, and peripheral (or other) compartments] that are interconnected in a physiologically relevant manner, have been developed in our laboratory (Appendix VI; Ramakrishnan et al., 2016). In this model, five subcompartments corresponding to Cyp27b1 enzyme in kidney and Cyp24a1 enzyme in liver, kidney, ileum, and brain were used to account for changes in the synthesis and degradation of the enzymes with 1,25(OH)2D3 treatment. Again, the PBPK-PD model incorporated FC in mRNA expression of both Cyp27b1 and Cyp24a1 (Fig. 7-2). To account for differences in intestinal blood/plasma flow to the enterocyte region of the intestine, we highlighted the SFM to contrast with the TM model (Fig. 7-2). These models were first introduced by Pang and colleagues (Cong et al., 2000; Doherty and Pang, 2000; Pang, 2003; Fan et al., 2010) to describe the differential perfusion rates to the small intestine. For the SFM, the intestine is viewed as two tissue subcompartments: the enterocyte compartment consists of absorptive/secretory transporters at the apical membrane facing the lumen, metabolic enzymes within, and a basolateral side facing the blood; while the serosal compartment acts only as a storage or distribution compartment. The fraction of intestinal blood flow perfusing the enterocyte region

(fQ) is 5-30% of the total intestinal blood flow (Cong et al., 2000; Pang and Chow, 2012). This model suggests that drug in the systemic circulation (e.g., from i.v. dosing) would be partially shunted away from the enterocyte region, whereas for p.o. dosing, the entire dose would first reach the enterocytes prior to entering the circulation. In contrast, the TM suggests that the entire intestinal blood flow perfuses the enterocyte region that is indistinguishable from the serosal region. Hence, the SFM suggests the occurrence of route-dependent intestinal metabolism, where a greater extent of intestinal metabolism occurs for p.o. or i.p. over i.v. dosing (Cong et al., 2000).

141

Figure 7-2. Schematic presentation of the PBPK-PD models for 1,25(OH)2D3 kinetics in mice. QT and VT denote the plasma flow rates and tissue volume, respectively, representing the plasma and tissue [brain (Br), kidney (K), liver (L), ileum (I), and peripheral (peri) or other tissue] compartments. fQ is the fractional intestinal (blood or plasma) flow perfusing the enterocyte region. CLint,met,T is the metabolic intrinsic clearance in tissue; ka and kdeg are the first-order absorption and degradation rate constants in the intestinal lumen, respectively. kin,Cyp24a1,T, kout,Cyp24a1,T, kin,Cyp27b1,K, and kout,Cyp27b1,K denote the turnover rate constants for Cyp24a1 and Cyp27b1, respectively, in subcompartments in various tissues and kidney. An assumption made was that the i.p. dose was absorbed solely by the intestine. See Tables 7-1 and 7-2 for detailed description of assigned and fixed parameters used in the models. For the SFM, the intestine was viewed as two tissue subcompartments (serosa and enterocytes) perfused by the serosal (70-95% total intestinal flow) and enterocyte (5-30% total intestinal flow) flow (right). This intestinal unit may be substituted into the PBPK(TM)-PD model to obtain the PBPK(SFM)-PD model.

142

Table 7-1. Assigned physiologic parameter values for PBPK-PD models

Parameter (Unit) Definition Value VP (mL) Plasma volume 0.962 VK (mL) Kidney volume 0.417 VL (mL) Liver volume 1.37 a VI (mL) Intestine volume 0.632 Ven = fQVI (mL) Enterocyte volume --- Vser = (1-fQ )VI (mL) Serosal volume --- VBr (mL) Brain volume 0.412 Vperi or Vothers (mL) Peripheral compartment volume 18.5 -1 b QK (mL·min ) Plasma flow to kidney 0.733 -1 b QL (mL·min ) Plasma flow to liver 1.3 -1 b QHA (mL·min ) hepatic arterial plasma flow rate 0.26 -1 a,b QI (mL·min ) Plasma flow to intestine 1.04 -1 Qen = fQQI (mL·min ) Plasma flow to enterocyte --- -1 Qser=(1-fQ )QI (mL·min ) Plasma flow to serosa --- -1 b QBr (mL·min ) Plasma flow to brain 0.266 -1 c QCO (mL·min ) Cardiac output 8.04 KL Partition coefficient of liver (liver to plasma concentration ratio) 0.15 KK Partition coefficient of kidney (kidney to plasma concentration ratio) 0.34 KBr Partition coefficient of brain (brain to plasma concentration ratio) 0.05 Partition coefficient of intestine, serosa, or enterocyte K = K = K 0.4 I ser en (intestine to plasma concentration ratio) -1 CP,baseline (pmol·kg ) Baseline plasma 1,25(OH)2D3 concentration 217 d -1 CK,baseline (pmol·kg ) Baseline renal tissue 1,25(OH)2D3 concentration 73.5 d -1 CL,baseline (pmol·kg ) Baseline liver tissue 1,25(OH)2D3 concentration 30.3 d -1 CBr,baseline (pmol·kg ) Baseline brain tissue 1,25(OH)2D3 concentration 10.8 d -1 CI,baseline (pmol·kg ) Baseline intestinal tissue 1,25(OH)2D3 concentration 86.6 Note: volume (V) and plasma flow (Q) were obtained from Davies and Morris (1993) and Brown et al. (1997); KT and plasma baseline concentrations, CP,baseline, were obtained experimentally (Chow et al., 2013b). a For TM, fQ = 1; for SFM, fQ was estimated from fitting; fQVI=Ven and (1-fQ)VI=Vser ; fQQI=Qen and (1-fQ)QI=Qser. b Values of tissue plasma flow were calculated as % plasma cardiac output (QCO) (Brown et al., 1997). c -1 Plasma cardiac output (QCO) for mice was calculated using an allometric relationship: QCO,mice (mL·min ) = 275x(1-Hct)x(body weight of mice in kg)0.75, where Hct is hematocrit (Brown et al., 1997). d CT,baseline was calculated according to CP,baseline and the apparent KT value (CT,baseline/CP,baseline) for kidney, liver, brain, and intestine 0.34, 0.14, 0.05, and 0.4 (Chow et al., 2013b).

143

7.3.3.2 The PBPK(SFM)-PD model was superior to the PBPK(TM)-PD model for describing the i.p. and i.v. data

Fitting with ADAPT5 was previously performed, and the results showed that the PBPK model with the SFM-nested intestine compartment was better in predicting the temporal data of

1,25(OH)2D3 and FC of Cyp27b1 and Cyp24a1 upon repeated i.p. dosing with 1,25(OH)2D3 [120 pmol, q2d x 4] (Appendix VI; Ramakrishnan et al., 2016). The PBPK(SFM)-PD model also proved to predict the i.v. data better. The fitted parameters obtained from both the TM- and SFM-nested PBPK-PD models (Ramakrishnan et al., 2016) are summarized in Table 7-2.

144

Table 7-2. Fitted parameters [estimate (CV%)] obtained from PBPK-PD models with TM and SFM nested within the model

Fitted Parameters Definition PBPK(TM)-PD PBPK(SFM)-PD fQ Fraction of intestinal flow to enterocyte region 1 0.105 (0.533) fd Fraction of cardiac output to peripheral compartment 0.0035 (0.002) 0.0048 (0.001) -1 ka (h ) Absorption rate constant of 1,25(OH)2D3 1.50 (0.004) 1.61 (0.0021) -1 kdeg (h ) Degradation rate constant of 1,25(OH)2D3 in lumen 0.0017 (2.34) 0.0012 (6.65)

Kperi or Kothers Partition coefficient of peripheral/other compartment 0.272 (0.003) 0.335 (0.003)

-1 Rsyn (fmol·h ) Endogenous synthesis rate of 1,25(OH)2D3 31.0 (0.247) 21.5 (0.116) kin,Cyp27b1,K or e Turnover rate constants of renal Cyp27b1 0.220 (11.5) 0.245 (9.72) kout,Cyp27b1,K Hill coefficient for indirect response of renal Cyp27b1 γ 3.57 (14.0) 2.71 (12.1) 2 function Hepatic metabolic intrinsic clearance of 1,25(OH) D via f CL (mL·h-1)a 2 3 0.0043 (1.78) 0.0010 (3.03) L int,met,L hepatic Cyp24a1 Intestinal metabolic intrinsic clearance of 1,25(OH) D via f CL (mL·h-1)a 2 3 0.0011 (0.038) 0.0014 (0.412) L int,met,I intestinal Cyp24a1 Renal metabolic intrinsic clearance of 1,25(OH) D via renal f CL (mL·h-1)a 2 3 0.0242 (0.008) 0.0280 (0.0066) L int,met,K Cyp24a1 Brain metabolic intrinsic clearance of 1,25(OH) D via brain f CL (mL·h-1)a 2 3 0.0006 (8.30) 0.0003 (0.412) L int,met,Br Cyp24a1 Hill coefficient for indirect response function of hepatic γ 1.24 (2.45) 1.75 (0.020) Cyp24a1 Hill coefficient for indirect response function of renal γ 2.64 (0.665) 3.59 (0.081) Cyp24a1 Hill coefficient for indirect response function of intestinal γ 0.985 (1.22) 2.09 (0.002) Cyp24a1 Hill coefficient for indirect response function of brain γ 0.878 (0.148) 0.576 (0.076) Cyp24a1 kin,Cyp24a1,L or e Turnover rate constant of hepatic Cyp24a1 0.045(0.006) 0.047 (0.008) kout,Cyp24a1,L kin,Cyp24a1,I or e Turnover rate constant of intestinal Cyp24a1 0.489 (0.005) 0.287 (0.067) kout,Cyp24a1,I kin,Cyp24a1,I or e Turnover rate constant of renal Cyp24a1 0.012 (0.065) 0.047 (0.008) kout,Cyp24a1,I kin,Cyp24a1,Br or e Turnover rate constant of brain Cyp24a1 1.28 (0.051) 1.87 (5.10) kout,Cyp24a1,Br AIC Akaike information criteria 10006 9993 WSSR Weighted sum of squared residuals 1664 1478 b df Degrees of freedom (Fcritical = 3.84) 986 985 F-test valuec Calculated F score 124d a Fitted CLint,met,T values were adjusted by corresponding unbound fraction of 1,25(OH)2D3 in tissue (fT). b Degrees of freedom is the number of data points (n = 1010) used in the model minus the number of parameters being fitted. c F score was calculated using × , where dfj > dfi. d F score suggests a significant improvement in goodness of fit (Fcritical = 3.84) for PBPK(SFM)-PD versus PBPK(TM)-PD (Boxenbaum et al., 1974). e Normally the units for the zero-order production (kin) and first-order degradation (kout) rate constants, are mass per unit time and 1/time, respectively. However, when the value is normalized to the baseline value, the rate constant becomes unitless.

145

7.3.3.3 The extended PBPK(SFM)-PD model

Because the PBPK(SFM)-PD model proved to be superior in predicting the disposition of

1,25(OH)2D3, we extended this model to incorporate the pharmacological actions of 1,25(OH)2D3 in the liver (Fig. 7-3). Fitting was performed with ADAPT5 with the established PBPK(SFM)-PD model with the assigned (Table 7-1) and fitted (Table 7-2) parameters to examine whether the

PBPK(SFM)-PD model was able to predict cholesterol lowering, where 1,25(OH)2D3 was previously found to induce both Cyp7a1 mRNA and protein expression to increase metabolism of liver cholesterol in mice in vivo (Chow et al., 2014).

Figure 7-3. Extension of the PBPK(SFM)-PD model (see Fig. 7-2) to incorporate 1,25(OH)2D3-mediated activation of liver Cyp7a1, leading to increased metabolism of cholesterol [data of Chow et al., (2014)]. Induction is represented by the open box; CL, clearance of 1,25(OH)2D3; Q, blood flow; kin,Cyp24a1, zero-order synthesis rate constant of Cyp24a1; kout,Cyp24a1, first-order degradation rate constant of Cyp24a1; kin,Cyp7a1, zero-order synthesis rate constant of Cyp7a1; kout,Cyp7a1, first-order degradation rate constant of Cyp7a1; kin,cholesterol, zero- order synthesis rate constant of cholesterol; kout,cholesterol, first-order degradation rate constant of cholesterol; τ, time delay function.

The extended PBPK(SFM)-PD model (Fig. 7-3) incorporated synthesis and degradation rate constants for Cyp7a1 and cholesterol. The FC in mRNA or protein expression of hepatic Cyp7a1 was used to predict % reduction of liver cholesterol following the repeated doses of i.p.

1,25(OH)2D3 treatment [120 pmol, q2d x 4] in mice fed the high fat/high cholesterol diet (data of Chow et al., 2014). We utilized the same strategy as Ramakrishnan and colleagues (2016), where

induction of Cyp7a1 after 1,25(OH)2D3 i.p. dosing was described with an indirect response model

which incorporates the full sigmoidal Emax, EC50, and Hill coefficients (Equation D1, Appendix D). These constants were first estimated by regression of the FC of Cyp7a1 against the hepatic

1,25(OH)2D3 concentration (Supplementary Fig. S7-1) and final estimates were obtained from 146

fitting. We incorporated transit compartments and a time delay function (Equation D2, Appendix D) to improve fitting to explain the Cyp7a1-mediated induction of cholesterol metabolism (Equation D3, Appendix D).

7.4 Results

7.4.1 Dose-dependent PK of the i.v. 1,25(OH)2D3 data according to the two- compartment model

1,25(OH)2D3 concentrations for the 60 and 120 pmol doses fell below basal values (187 ± 48.5 pM) by 24 h whereas those for the 2 pmol dose remained relatively unchanged. Plasma

1,25(OH)2D3 concentrations decayed biexponentially at each dose level (Fig. 7-4). The t1/2β (36.7

h) was prolonged for the 2 pmol dose but dramatically shorter t1/2βs (~6 h) existed for higher doses

(Table 7-3). Non-compartmental values of AUC∞/dose decreased with increasing dose, yielding -1 -1 greater clearance values at higher doses. CLtotal increased from 0.1 to 2.0 mL·min ·kg (Table 7-

3), an observation compatible with induction of Cyp24a1 for the metabolism of 1,25(OH)2D3.

Table 7-3. Non-compartmental pharmacokinetic parameters following administration of first i.v. doses of 1,25(OH)2D3 to mice Dose (pmol)

2 60 120 a t1/2β (h) 36.7 6.6 6.0 b AUC∞ (pM·h) 14900 24500 57200 -1 AUC∞/dose (h·L ) 7460 409 477 -1 -1 c CLtotal (mL·min ·kg ) 0.112 2.04 1.75 a t1/2β was calculated as 0.693/β, where β is the terminal slope of ln(concentration) versus time data between 6-48 h. b AUC∞ was determined as (AUC0→48 + Clast/β), and AUC0→48 was estimated by the trapezoidal rule. c CLtotal was estimated as dose/AUC∞.

147

Figure 7-4. Plasma 1,25(OH)2D3 concentrations after i.v. administration of 2, 60 and 120 pmol doses versus basal levels. The two-compartment model (Fig. 7-1) was used to fit the 1,25(OH)2D3 data for each single 2, 60 or 120 pmol dose (A) individually, (B) combined fit of data for the first doses, with unscaled or scaled k12, and (C) combined fit of data for all doses, with unscaled or scaled k12. Observed plasma 1,25(OH)2D3 concentrations are shown as mean ± SEM (circle, n = 3-4 different mice) with fitted values shown as a solid line. Data for vehicle-treated mice (basal level) were averaged and joined by the dashed line (n = 9).

148

For the simple two-compartment model that contained Rsyn for 1,25(OH)2D3 formation from its

vitamin D precursors, fits to the single dose data individually revealed a decreasing Rsyn and an increasing k10 with increasing dose (Fig. 7-4A; Table 7-3). Furthermore, fitted values of k12 and

Vss increased with dose, suggesting a larger distribution volume with increasing dose levels. These distributional changes could not be explained by a saturation of protein binding sites as the 5 µM concentration of DBP in plasma (Chun, 2012) greatly exceeds the plasma 1,25(OH)2D3 concentrations from i.v. dosing. We then performed combined fitting of all data from the first doses and compared the parameter estimates obtained to those from individual fits.

Accommodation of the changing k12 was accomplished by scaling k12 with dose (Table 7-3;

Appendix A). The model predicted the data for the first dose well, whether or not k12 was scaled

(Fig. 7-4B). For the two-compartment model, the fitted k12 value for the 2 pmol dose (2.77 ± 1.42 h-1), obtained from the individual fit of the 2 pmol data (Fig. 7-4A), was about half that from forced

-1 fitting of data from the first doses (4.15 ± 1.23 h ) and with scaling of k12 (Table 7-3). For fitting of the first and repeated doses, the fit to the third 120 pmol dose was better than the fit to the first

dose, with or without k12 scaled (Fig. 7-4C). For the forced fit to all data from single and repeated

dosing, the averaged value of the fitted k12 was halved when k12 was scaled, k21 and k10 were lower and V1 and Vss were higher (Table 7-3). The WSSR was smaller with scaled k12 (although the

AICs were similar), suggesting that scaling of k12 was an improvement (Table 7-3). The lack of a

significant improvement in the forced fit for scaled k12 versus unscaled k12 was likely due to

inability of this model to account for the dose-dependent nature of Rsyn and k10.

149

Table 7-4. Fitted pharmacokinetic parameters for first i.v. doses of 1,25(OH)2D3 according to the simple two-compartment model Individual fitting of first dose data Forced fitting of first doses 2, 60 and 120 pmol 2, 60 and 120 pmol 2 pmol 60 pmol 120 pmol k12 not scaled k12 scaled -1 c k12 (h ) 2.77 ± 1.42 4.17 ± 3.57 15.8 ± 4.36 10.8 ± 4.67 4.15 ± 1.23 -1 k21 (h ) 2.15 ± 1.14 1.41 ± 0.74 2.09 ± 0.40 2.93 ± 0.10 1.89 ± 0.49 -1 k10 (h ) 0.27 ± 1.10 0.93 ± 0.44 2.21 ± 0.71 1.11 ± 0.71 0.75 ± 0.37 -1 V1 (mL·kg ) 112 ± 17.0 167 ± 81.5 61.5 ± 21.0 72.0 ± 1.6 117 ± 10.5 -1 a Vss (mL·kg ) 255 659 526 337 409 -1 b Rsyn (fmol·h ) 71.9 ± 39.8 31.1 ± 10.8 32.6 ± 5.91 85.5 ± 4.78 81.6 ± 16.4 WSSR 9.05 5.73 AIC 494 523 a Vss was calculated as V1+V2, where V2, peripheral compartment = V1·(1+k12/k21). b -1 Rsyn initial estimate (49.6 fmol·h ) was obtained from Hsu et al. (1987). c SD of parameter estimate.

7.4.2 Cyp27b1 and Cyp24a1 mRNA expression for the i.v. data

The mRNA expression of the synthetic enzyme in kidney and degradation enzyme in intestine and kidney also displayed dose-dependent changes (Fig. 7-5). For the 2 pmol single dose, there was an absence of any notable trend for the mRNA expression of renal Cyp27b1 or renal and ileal Cyp24a1, as levels remained relatively unchanged in relation to basal values (Fig. 7-5, left).

Furthermore, hypercalcemia was not observed for the low dose (data not shown). Thus, the Rsyn value obtained for the 2 pmol dose should represent the net synthesis rate of endogenous

1,25(OH)2D3 formation from its vitamin D precursors. In contrast, markedly lower Cyp27b1

mRNA expression at 9 h after 1,25(OH)2D3 administration was noted with higher doses, wherein levels fell and remained below basal levels. Maximal induction of renal Cyp24a1 expression occurred at 6-9 h and was sustained until 48 h. In ileum, maximal induction of Cyp24a1 mRNA expression occurred at 3 h following the 60 pmol dose and at 6 h following the 120 pmol dose

(Fig. 7-5, left). Repeated administration of 120 pmol 1,25(OH)2D3 led to a greater downregulation (Cyp27b1) and induction (Cyp24a1) of renal mRNA levels when compared with the single dose and a similar pattern was observed for ileal Cyp24a1 mRNA expression (Fig. 7-5, right).

150

Figure 7-5. Relative mRNA expression for synthesis and degradation enzymes following single or repeated i.v. administration of 1,25(OH)2D3. (A) Renal Cyp27b1 mRNA expression is reduced by both single and repeated administration of 60 and 120 pmol 1,25(OH)2D3. (B) Renal and (C) ileal Cyp24a1 mRNA expression are induced by 1,25(OH)2D3 in a dose-dependent manner for single and repeated dosing. Data for vehicle-treated mice (basal level) were averaged and joined by the dashed line (n = 9), whereas data for treated mice are mean ± SEM and joined by a solid line (n = 3-4 different mice). Significant differences between groups were denoted by: a, basal level versus 2 pmol; b, basal level versus 60 pmol; c, basal level versus 120 pmol; †, 2 pmol versus 60 pmol; ‡, 2 pmol versus 120 pmol; #, 60 pmol versus 120 pmol; *, all groups except basal level versus 2 pmol.

151

7.4.3 PD response versus concentration curves for the i.v. data

Fig. 7-6 shows the concentration-response relationship for FC of renal Cyp27b1 and Cyp24a1 and intestinal Cyp24a1 mRNA expression in mice receiving single (2, 60 and 120 pmol) and repeated

(2 and 120 pmol) doses of 1,25(OH)2D3. A plateau was reached for Cyp27b1 and Cyp24a1 within the dose-range and FC for Cyp27b1 downregulation and Cyp24a1 induction remained constant at

1,25(OH)2D3 concentrations >5000 pM. Upon fitting of Equation 1, a threefold increase in Imax

was observed after repeated dosing (Table 7-5; Fig. 7-6A), although the fitted IC50 values were

similar for the single and repeated doses. Composite Imax and IC50 values were also obtained upon regression of pooled data from the first and repeated doses with Equation 1, and these values were

similar to those obtained for the repeated third dose (Table 7-5). Estimated Emax values (expressed over basal level) obtained for Cyp24a1 (Equation 2) for the first and the third doses were virtually

identical (Figs. 7-6, B and C), whereas EC50 values for both renal and intestinal Cyp24a1 increased significantly upon repeated dosing compared with that for the first dose (Table 7-5). Values for

Emax and EC50 for the pooled data were similar to those for the third dose. For fitting of Equation

B1 of Appendix B, the Emax values obtained from the pooled data for renal and ileal Cyp24a1 were

summed to provide the total Emax; the EC50 was estimated as the average of the EC50 values for the renal and intestinal Cyp24a1.

Table 7-5. Pharmacodynamic parameters estimated from Cyp27b1 or Cyp24a1 fold change versus plasma 1,25(OH)2D3 concentration

Renal Cyp27b1 Renal Cyp24a1 Ileal Cyp24a1 a a a Imax IC50 (pM) Emax EC50 (pM) Emax EC50 (pM) For first dose 8 ± 2b 100 ± 9 82 ± 19 300 ± 23 1000 ± 129 500 ± 62 For repeated third dose 24 ± 2 150 ± 15 107 ± 16 1480± 90 939 ± 126 3630 ± 220 For first and third doses 24 ± 4 150 ± 24 107 ± 19 1480 ± 189 970 ± 218 3630 ± 272 a Imax and Emax values are expressed as FC relative to baseline. b SD of parameter estimate.

152

Figure 7-6. Inhibition of Cyp27b1 and induction of Cyp24a1 by 1,25(OH)2D3. Plots of renal Cyp27b1 and renal and intestinal Cyp24a1 mRNA FC versus plasma concentration of 1,25(OH)2D3 following first (circle) or third (square) dose of 2 (red), 60 (blue) and 120 (green) IC ECmax p pmol, with equations for inhibition Cyp27b1 =(1- max p ) and induction Cyp24a1 =(1+ ) . FC FC EC +C IC50 +C p 50 p

153

7.4.4 Fitting of the PKPD model to i.v. 1,25(OH)2D3 concentration

Owing to the dose discrepancy and progressive changes that occur for Rsyn and k10 with dose (Table 7-4), there was a need to incorporate the PD changes of Cyp27b1 and Cyp24a1 into the two- compartment model shown in Fig. 7-1A (Appendix B). Hence, a PKPD model (Fig. 7-1B) was

developed using scaled k12 values for greater doses. From fitting with data from the first doses, the

EC50 (average of renal and ileal Cyp24a1 EC50) and the summed Emax for the kidney and intestine were used as initial estimates. Additionally, scaling factors and Hill coefficients for the inhibition and induction functions were added stepwise to monitor improvement of fit in order to determine the best predictive model. For data from the single doses, forced fitting with the PKPD model showed significant improvement compared with the simple two-compartment model (Fig. 7-7A and Table 7-6). The best outcome was obtained when all of these modifications were incorporated into the model (Table 7-6). Values of the Hill coefficients were close to unity, Rsyn was 61.5 ± 5.24 -1 fmol·h (a value lower than that estimated from the two-compartment model), V1 was similar to

plasma volume and k10 [representing the basal elimination rate constant of 1,25(OH)2D3] was 0.128 ± 0.021 h-1, a value lower than but reasonable to that estimated from the simple two- compartment model.

For fitting of data from all doses (Table 7-6), we further adopted a second strategy as there were

changes in the EC50 upon repeated dosing. We used one EC50 [obtained from regression of FC vs.

1,25(OH)2D3 plasma concentrations from the first dose] or two different EC50 values [EC50(1) from

regression of data from first dose and EC50(2) from pooled data]. When using one EC50, it was

assumed that the EC50 for the first dose was identical to that for the second and third doses and the regressed value from data from the first dose was used as the initial estimate. When using different

EC50 values, EC50(1) obtained from the first dose and EC50(2) from pooled data were used as initial

estimates. These EC50 values, together with scaled k12 functions, scaling factors, and/or Hill coefficients in the PKPD model, were incorporated in a stepwise manner for fitting (data not

shown). From fits based on one or two EC50 values, we found that the parameters (k12, k21, k10, V1 and Rsyn) and the scaling factors remained similar. The Hill coefficients using two EC50 values remained similar to unity, whereas those for the one EC50 model were greater (Table 7-6). Overall, it was concluded that the best fit was attained when all of these modifications, including different

EC50 values, were added to the PKPD model (Table 7-6). We also used parameters obtained from 154

the 2 pmol first dose to simulate plasma 1,25(OH)2D3 concentration-time data for other single and

repeated dosing regimens, with use of one EC50 (Fig. 7-7C) or two different EC50s (Fig. 7-7E). The simulations predicted the first dose reasonably well, but were unable to fully capture the later concentrations upon repeated dosing, presumably due to tolerance.

155

Figure 7-7. Fitting of single and repeated dose plasma 1,25(OH)2D3 i.v. data using a PKPD model. (A) Simultaneous fitting of single dose plasma 1,25(OH)2D3 data. Simultaneous fitting of combined single and repeated dose plasma 1,25(OH)2D3 using (B) one EC50 and (C) simulations to predict data of higher doses using parameters from the 2 pmol dose. Simultaneous fitting with (D) different EC50s and (E) simulations to predict data of higher doses. Observed plasma 1,25(OH)2D3 concentrations are shown as mean ± SEM (n = 3-4) with fitted values shown as a solid line and simulated values shown as a dashed line. Data for vehicle-treated mice (basal level) were averaged and joined by the dashed line (n = 9).

156

Table 7-6. Simultaneous fitting of first dose or pooled first and third dose i.v. data to different models, with k12 scaled

Combined Fit to Data of First Dose Combined Fit to Data of All Doses (First and Repeated Third Dose) Fitted Parameters Two-Compartment PKPD Indirect Response Two-Compartment PKPD Model PKPD Model Indirect Response Model Model Model Model (one EC50) (different EC50) Model -1 a d k12 (h ) 4.15 ± 1.23 4.71 ± 2.22 3.88 ± 3.52 3.23 ± 0.77 2.75 ± 1.42 2.74 ± 1.34 6.17 ± 4.81 -1 k21 (h ) 1.89 ± 0.49 3.01 ± 0.24 1.70 ± 1.21 2.22 ± 0.55 2.83 ± 1.40 3.25 ± 1.42 3.75 ± 2.68 -1 k10 (h ) 0.75 ± 0.37 0.128 ± 0.021 0.349 ± 0.096 0.55 ± 0.27 0.091 ± 0.034 0.092 ± 0.007 0.133 ± 0.118

V1 (mL) 2.33 ± 0.21 1.38 ± 0.30 1.79 ± 0.55 2.43 ± 0.31 1.86 ± 0.49 1.86 ± 0.49 1.43 ± 0.83 -1 Rsyn (fmol·h ) 81.6 ± 16.4 61.5 ± 5.24 49.1 ± 9.2 78.9 ± 42.8 47.2 ± 16.1 44.5 ± 6.47 29.4 ± 19.0

Emax 398 ± 54.6 802 ± 175 404 ± 80.7 407 ± 47.4 1050 ± 515

EC50(1) (pM) 713 ± 137 1530 ± 1280 790 ± 248 616 ± 296 3110 ± 2990

EC50(2) (pM) 3810 ± 326 4640 ± 6110

Imax 11.3 ± 3.30 23.4 ± 5.41 11.6 ± 2.56 10.0 ± 3.07 45.2 ± 14.6

IC50 (pM) 276 ± 91.0 143 ± 218 306 ± 49.7 310 ± 74.8 388 ± 701

SM1 4.24 ± 0.42 4.47 ± 0.87 5.41 ± 0.81 7.15 ± 1.38 0.81 ± 0.52

SM2 25.8 ± 2.87 75.4 ± 16.6 25.9 ± 4.66 26.8 ± 6.20 9.64 ± 10.4

γ1 1.25 ± 0.44 2.80 ± 0.96 1.85 ± 1.82 0.783 ± 0.391 2.12 ± 0.49

γ2 0.983 ± 0.249 2.14 ± 0.21 1.33 ± 0.16 1.15 ± 0.154 2.50 ± 0.53 -1 b,c kin1 (h ) 0.390 ± 0.163 0.343 ± 0.083 -1 b,c kin2 (h ) 0.195 ± 0.110 0.362 ± 0.175 τ (h) 0.809 ± 0.114 1.36 ± 1.07 WSSR 5.73 0.47 709 30.2 23.6 11.8 676 AIC 523 649 982 913 1067 1086 1547 F-test value 28.3* -1.53 1.48 7.09* -1.32

Fcritical (df1, df2) 2.45 (8, 20) 2.41 (11, 17) 2.17 (8, 42) 2.12 (9, 41) 2.43 (12, 16) a Values of k12 for the 60 and 120 pmol doses were assigned as 1.5x and 5.7x the value of k12 for the 2 pmol dose. b kin1 and kin2 are the zero-order synthesis rate constants for Cyp27b1 and Cyp24a1, respectively. c kout1 and kout2 are the first-order degradation rate constants for Cyp27b1 and Cyp24a1, respectively, and are equal to kin1*baseline or kin2*baseline, where baseline is 1. d SD of parameter estimate. *Significant difference between PKPD versus two-compartment model at P < 0.05 compared with Fcritical.

157

7.4.5 Indirect response model with inhibition and induction functions to explain i.v. 1,25(OH)2D3 data with the compartmental PKPD model

The PD components of the indirect response model incorporated both an indirect inhibitory (Cyp27b1) and stimulatory (Cyp24a1) response model, and transit compartments with lag time (τ) were added to explain the time delay for Cyp27b1 effects since Cyp27b1 was indirectly affected

by VDR activation through PTH (Fig. 7-1C). A zero-order rate constant for formation (kin), a first- order decay rate constant (kout) and the appropriate inhibition and induction functions were added to account for decreased Cyp27b1 and increased Cyp24a1 production (Appendix C) (Dayneka et al., 1993; Sharma and Jusko, 1996). Preliminary fits showed that two additional transit compartments were needed to improve model fitting (data not shown). Improved fits were obtained when scaling factors and Hill coefficients were incorporated into the model. For the single dose data, the fitted parameter values for k12, k21, k10, V1 and Rsyn were usually within twofold (Table

7-6). For the pooled data, when different EC50 values were used, greater values for k12 and EC50(1) values were obtained, whereas Rsyn and the scaling factors were smaller (Table 7-6). Model fitting criteria were not statistically improved compared to the two-compartment model, although the

indirect response model was able to reveal correlations between PD responses to 1,25(OH)2D3 concentrations in a temporal fashion (Fig. 7-8).

7.4.6 Comparison of the two-compartment, PKPD, and indirect response models

Fitting of the two-compartment model to individual i.v. data sets revealed dose-dependent kinetics and reasonably good fits (Fig. 7-4). However, fitting with the PKPD model for the single dose data proved to be superior (Fig. 7-7; Table 7-6). For repeated dosing, we found that the PKPD model

with different EC50 values was associated with smaller WSSR values with a significant F-value compared to the two-compartment model (P < 0.05), despite the larger AIC (Fig. 7-7D; Table 7-

5). Furthermore, incorporation of scaled k12 constants, scaling factors and Hill coefficients into the PKPD model improved model performance (data not shown). Although the indirect response model did not improve the fit statistically compared with the two-compartment model, the model

158

showed reasonable utility and was adequate in predicting changes in the temporal PD profiles against 1,25(OH)2D3 concentrations at the different administered dose levels (Fig. 7-8).

Figure 7-8. The indirect response model for simultaneous fitting of all i.v. data on plasma 1,25(OH)2D3 and FC of VDR target genes, Cyp27b1 and Cyp24a1. (A) Simultaneous fitting of single dose (left) and single and repeated dose (right) plasma 1,25(OH)2D3 concentration. Simultaneous fitting for (B) inhibition of Cyp27b1 FC and (C) stimulation of Cyp24a1 FC. Observed plasma 1,25(OH)2D3 concentrations and Cyp27b1 and Cyp24a1 FC are shown as mean ± SEM (n = 3-4 different mice) with fitted values shown as a solid line. Data for vehicle-treated mice (basal level) were averaged and joined by the dashed line (n = 9).

159

7.4.7 Use of PBPK(TM)-PD and PBPK(SFM)-PD models to simulate i.v. data

Simulations were performed for the escalating i.v. doses (2, 60, and 120 pmol) with the published PBPK(TM)-PD and PBPK(SFM)-PD models (Ramakrishnan et al., 2016). The models were able to predict the i.v. data of 1,25(OH)2D3 following different doses and repeated administration (Fig.

7-9). Of note was that the observed and predicted Cmaxs in ileum following i.v. administration were unexpectedly lower than the Cmaxs after i.p. administration (Ramakrishnan et al., 2016) despite the bioavailability of 0.84 ± 0.16 [from the dose-corrected area under the curve ratios of i.p./i.v.,

(AUCi.p./Dosei.p.)/(AUCi.v./Dosei.v.)], suggesting a lesser distribution of 1,25(OH)2D3 into the enterocyte after i.v. dosing, a flow pattern lending support to the SFM model. The PBPK(SFM)-

PD model was superior for describing the route-dependent kinetics of 1,25(OH)2D3 for i.p. dosing than the PBPK(TM)-PD model as it captured the lower Cmax in ileum following i.v. administration (Fig. 7-9).

Figure 7-9. Observed (closed circles) versus simulated (spline lines) concentration-time profiles of 1,25(OH)2D3 after multiple i.v. doses (given every 2 days for 6 days) in plasma, liver, ileum, and kidney using the PBPK-PD models with nested TM (dashed line) and SFM (solid lines) for describing the intestine compartment. Data were simulated with PBPK-PD using parameters in Table 7-2. In particular, Cmax 1,25(OH)2D3 values in ileum were much overestimated for the TM.

160

7.4.8 Extending the PBPK(SFM)-PD model to the liver

The fit to the temporal Cyp7a1 and liver cholesterol profiles following repeated i.p. doses of

1,25(OH)2D3 are shown in Figs. 7-10 and 7-11. The temporal Cyp7a1 data was from mice fed normal diets and sacrificed at several time points during the treatment period [120 pmol, q2d x4], whereas the liver cholesterol levels were obtained from mice fed a high fat/high cholesterol diet for 3 weeks and sacrificed 50 h after the last dose [120 pmol, q2d x4 during the last week of diet] (both sets of data are from Chow et al., 2014). The model assumes that there is no difference in basal Cyp7a1 expression between normal vs. high fat/high cholesterol diets, as previously reported

(Chow et al., 2014). The % reduction in liver cholesterol was obtained from 1,25(OH)2D3-treated mice normalized to vehicle-treated mice on high fat/high cholesterol diet, where the vehicle-treated mice are defined as the initial condition (100%). Simulations were performed to describe untreated conditions (dashed line). The observed (circles) and model-fitted (solid line) data were similar, showing that the model was able to adequately describe the induction pattern of Cyp7a1 mRNA, although the high expression of Cyp7a1 upon repeated dosing was not fully captured (Fig. 7-10A). The extended PBPK(SFM)-PD model was also able to describe the temporal Cyp7a1 protein data well (Fig. 7-11A). When compared with using the Cyp7a1 mRNA data for fitting, the CV% for fitted parameters and AIC value were much smaller with Cyp7a1 protein data (Supplementary Table S7-1). This discrepancy was likely a result of the large variation in the temporal mRNA data, especially upon repeated dosing where the mRNA data could not be fully captured with the extended PBPK(SFM)-PD model, contrasted by the more uniform pattern that was observed with Cyp7a1 protein (Figs. 7-10A and 7-11A). However, values from the F-test confirmed that the extended PBPK(SFM)-PD model was not significantly altered with the use of protein data (Supplementary Table S7-1). Because Cyp7a1 expression is influenced by circadian rhythm (Noshiro et al., 1990; Chow et al., 2014), the mass balance equations could potentially be modified to describe diurnal variation using simple cosine functions (Appendix D). However, due to the complexity of the underlying mechanisms controlling the circadian rhythm of Cyp7a1, including its regulation by core clock genes (Ferrell and Chiang, 2015), the time of feeding, and circadian rhythm of cholesterol synthesis by 3-hydroxy-3-methylglutaryl coenzyme A reductase (HMGCR) (Cella et al., 1995), these simple equations could not adequately describe the diurnal variation of Cyp7a1 (data not shown).

161

Figure 7-10. The extended PBPK(SFM)-PD model could adequately describe temporal Cyp7a1 mRNA and cholesterol levels in the liver [data of Chow et al., (2014)]. The untreated controls were simulated (dashed line). (A) The observed (circle) and fitted (solid line) time course of Cyp7a1FC mRNA after multiple i.p. doses using the base PBPK(SFM)-PD model described a (B) reduction in liver cholesterol upon treatment with 1,25(OH)2D3. Note: temporal Cyp7a1 mRNA data are from mice fed a normal diet and sacrificed throughout the treatment period, whereas liver cholesterol levels are from mice fed a high fat/high cholesterol diet for 3 weeks and sacrificed at the end of the treatment period.

Figure 7-11. The extended PBPK(SFM)-PD model could also adequately describe temporal Cyp7a1 protein and cholesterol levels in the liver [data of Chow et al., (2014)]. The untreated controls were simulated (dashed line). (A) The observed (circle) and fitted (solid line) time course of Cyp7a1FC protein after multiple i.p. doses using the base PBPK(SFM)-PD model described a (B) reduction in liver cholesterol upon treatment with 1,25(OH)2D3. Note: temporal Cyp7a1 protein data are from mice fed a normal diet and sacrificed throughout the treatment period, whereas liver cholesterol levels are from mice fed a high fat/high cholesterol diet for 3 weeks and sacrificed at the end of the treatment period.

162

7.5 Discussion

1,25(OH)2D3 is used extensively for the treatment of secondary hyperparathyroidism in uremic patients (Slatopolsky et al., 1984; Kimura et al., 1991; Brandi et al., 2002) and has demonstrated therapeutic potential for anticancer therapy (Hershberger et al., 2002; Rassnick et al., 2008;

Ramnath et al., 2013). The therapeutic use of 1,25(OH)2D3 is often limited by the propensity of

1,25(OH)2D3 to cause hypercalcemia and adverse effects. Recent studies have suggested that

1,25(OH)2D3 may also play a beneficial role in lowering cholesterol (Chow et al., 2014) as well as enhancement of amyloid-β efflux and reduction of cerebral plaque in transgenic mice expressing the human amyloid-β precursor protein (Durk et al., 2014). Thus, an understanding of the PK and

PD of 1,25(OH)2D3 is critical for the prediction of a proper dose and dosing regimen for potential therapeutic uses.

Information on the PK of 1,25(OH)2D3 is equivocal. Some of the discrepancies in the reported PK

parameters of 1,25(OH)2D3 may exist due to species differences, inadequate sampling or dose- dependent PK and altered PD with dose and route. An examination of available literature regarding

the PK properties of 1,25(OH)2D3 shows substantially different t1/2 values with respect to dose and

route of administration among species. In humans, a 4 µg dose of 1,25(OH)2D3 administered i.v.

or p.o. resulted in a t1/2 of 25.9 and 28.2 h, respectively (Brandi et al., 2002), whereas a similar i.v. -1 - dose of 0.06 µg·kg led to a shorter t1/2 of 16.5 h (Salusky et al., 1990). An equal dose of 20 µg·kg 1 administered i.p. and p.o. to the rat resulted in different t1/2 values of 5.0 and 10.4 h (Vieth et al., -1 1990a). Administration of 10 and 50 µg·kg of 1,25(OH)2D3 i.v. resulted in t1/2 of 3.8 and 2.3 h

respectively (Kissmeyer and Binderup, 1991). These shorter t1/2 values could be due to greater clearances, and are the consequence of a greater induction of Cyp24a1. In contrast, mice treated

with single doses of 0.125 or 0.5 µg of 1,25(OH)2D3/mouse i.p. did not exhibit a clear dose-

dependency, with t1/2 of 7.6 and 7.8 h (Muindi et al., 2004). Chow et al. (2013b) also reported a

t1/2 of 6.8 h in mice treated i.p. with repeated doses of 0.05 µg per mouse 1,25(OH)2D3.

Our present study with intravenous doses of 2, 60 and 120 pmol (0.00083, 0.025 and 0.05 µg per

mouse) 1,25(OH)2D3 clearly revealed dose-dependent trends in t1/2β values ranging from 36.7 to -1 -1 6.6 h (Table 7-3). Accordingly, CLtotal values of 0.1, 2.0 and 1.8 mL·min ·kg for 2, 60 and 120 pmol doses of 1,25(OH)2D3 were found (Table 7-3). Changes in enzymes were virtually absent for

163

the 2 pmol dose, rendering the fitted parameters according to the two-compartment model pertinent to basal conditions, whereas maximal inhibitory (Cyp27b1) and stimulatory (Cyp24a1) responses were observed for the higher doses. Important information was revealed from fitted results from

the lowest dose. For example, the volume of distribution of 1,25(OH)2D3 was low and Rsyn was 71.9 ± 39.8 fmol·h-1, a basal value that is higher than that (49.6 fmol·h-1) previously reported by

Hsu et al. (1987). Both the 60 and 120 pmol doses produced greater CLtotal values (2.0 and 1.8 -1 -1 mL·min ·kg ) that were similar to i.p. mouse studies, with apparent CLtotal values (CLtotal/F where F is bioavailability of i.p. dose) of 1.2 and 2.8 mL·min-1·kg-1, estimated graphically from the data of Muindi et al. (2004), for doses of 0.125 and 0.5 µg per mouse.

Gene profiling of Cyp27b1 and Cyp24a1 in the kidney and intestine, two major VDR-containing

tissues, confirmed the inhibition and induction associated with 1,25(OH)2D3 administration (Fig. 7-4). The dose-dependent downregulation of renal Cyp27b1 mRNA expression readily explains

the decreased Rsyn of 1,25(OH)2D3 for higher doses. Moreover, a dose-dependent induction of

renal and ileal Cyp24a1 mRNA expression further explained the increased CLtotal and decreased

t1/2 with increasing dose. Interestingly, the concentration-response curve for Cyp24a1 suggests

tolerance as EC50 values were increased in mice given repeated doses of 1,25(OH)2D3 without a

change in Emax (Fig. 7-6). The observed tolerance may be attributed to a decreased binding affinity

of 1,25(OH)2D3 to the VDR. We also observed a sustained induction of renal Cyp24a1 mRNA expression at 48 h, whereas ileal Cyp24a1 expression returned to basal levels by 24 h post-injection

for both single and repeated dosing. Long-term treatment with 1,25(OH)2D3 has been shown to lower Cyp24a1 mRNA expression, suggesting a lack of activation of VDR upon repeated dosing (unpublished data). Accordingly, the change in intestinal Cyp24a1 mRNA level in response to

1,25(OH)2D3 has been described to be more short-lived than in kidney and becomes refractory to

the continued administration of 1,25(OH)2D3, suggesting the presence of intestinal adaptation mechanisms which down-regulate the responsiveness of the enzyme (Lemay et al., 1995). The sustained induction of renal and not intestinal Cyp24a1, and the fact that little or no intestinal removal occurs for systemically administered compounds (due to shunting of intestinal flow away from the enterocyte region) (Doherty and Pang, 1997; Cong et al., 2000) suggest that renal

Cyp24a1 is the major enzyme for the catabolism of 1,25(OH)2D3 given i.v. Collectively, tolerance

of Cyp24a1 to repeated 1,25(OH)2D3 administration explains the inconsistency of parameters

164

derived from the PKPD model for the 2 pmol dose in predicting 1,25(OH)2D3 profiles for higher doses and upon repeated dosing.

This comprehensive analysis of 1,25(OH)2D3 quantification, together with gene profiling and data fitting using the PKPD and indirect response models, brings a new perspective as to how

1,25(OH)2D3 levels affect VDR target genes and how these dynamic changes affect the kinetics of

exogenously administered 1,25(OH)2D3. We recommend that PK studies involving 1,25(OH)2D3 incorporate PD effects on Rsyn and clearance in order to fully capture the disposition of

1,25(OH)2D3 since changes in Cyp27b1 and Cyp24a1 expression alter the synthesis and

degradation of 1,25(OH)2D3, events that can be explained with appropriate PKPD models. By incorporating the PD changes, we were able to integrate the gene changes that affect 1,25(OH)2D3 concentrations, showing the mutual interaction of kinetics and dynamics. We caution against merely reporting t1/2, especially when the data do not cover the time course describing the decay.

Our findings could explain the array of reported t1/2 with different doses in the literature, as we show that clearance is dose-dependent and that Cyp27b1 and Cyp24a1 work closely to affect

1,25(OH)2D3 levels and vice versa. Clearly, the simple two-compartment model, although

adequately showing dose-dependency in k12, Vss, Rsyn and CLtotal, could not fully explain the

intricacies of the dose-dependent PK of 1,25(OH)2D3. Using the derived parameters of Emax, EC50,

Imax and IC50, with or without incorporating changes in EC50 for single and repeated doses, we

could adequately predict the dose-dependent PKPD profiles of 1,25(OH)2D3. Further modifications to the model, including scaled k12, scaling factors and Hill coefficients significantly improved model performance. Although there is no apparent improvement with the indirect

response model statistically, we could demonstrate the direct correlation between 1,25(OH)2D3 and VDR target genes and essential time delays using transit compartments in the model (Mager and Jusko, 2001; Mager et al., 2003).

When compared with compartmental models, physiologically-based models are more useful for describing the disposition of 1,25(OH)2D3 as they further account for the synthesis and degradation

of 1,25(OH)2D3 in tissues. The finer mechanistic details can be obtained by applying PBPK models that are parameterized based on physiologic system components, functions, and tissue compartments that are connected by plasma and blood flow rates. PBPK-PD modeling can also be used to explain changes with respect to different routes of administration or patterns of intestinal

165

flow, and these models could be extended to incorporate PD effects such as those that pertain to the liver to describe the regulatory events of the VDR on cholesterol lowering genes. Here, the

modeling design was based on measurements of 1,25(OH)2D3 concentrations and regulatory

enzymes that are under feedback control by 1,25(OH)2D3 bound VDR over a sufficiently long duration following multiple i.p. and i.v. doses. We revisited the rich i.p. data from chapter 3 (Chow et al., 2013b) that included a dense sampling frequency and rich temporal gene profiles. Once these PBPK-PD models were established, the models were then used to predict the i.v. data of

1,25(OH)2D3 in mice presented earlier in this chapter (Quach et al., 2015). Our recently developed PBPK-PD model (Ramakrishnan et al., 2016) has provided a more complete description of the

kinetics of 1,25(OH)2D3 by including Cyp27b1 for the synthesis of 1,25(OH)2D3 in kidney and Cyp24a1 for catabolism in kidney, ileum, liver, and brain. Expectedly, inclusion of the inhibitory

effect of 1,25(OH)2D3 on Cyp27b1-mediated endogenous synthesis (kin) of 1,25(OH)2D3 in

kidney, which normally accounts for the bulk of circulating 1,25(OH)2D3 (Bell, 1998), aptly

described the 1,25(OH)2D3 profiles and the Cyp24a1 and Cyp27b1 expression. When comparing the SFM versus the TM, the SFM intestinal model was found to define the nonvascular (oral or i.p.) route of administration in more physiologically meaningful terms (Ramakrishnan et al., 2016) by distinguishing the differences in plasma/blood flow, transporter, channel, and metabolic enzyme density in the enterocyte versus serosal regions of the intestine (Cong et al., 2000). With the peritoneal cavity serving as a reservoir for i.p. dosing, passive, nonsaturable, and continuous absorption of 1,25(OH)2D3 into the enterocyte compartment ensued (Hollander et al., 1978). With

1,25(OH)2D3 given i.p., part of the dose must traverse the enterocyte layer before reaching systemic circulation, whereas with i.v. administration, the entire dose directly enters the circulation. Consequently, more 1,25(OH)2D3 is available for intestinal metabolism following i.p. or p.o. than i.v. administration (Cong et al., 2000), allowing a greater extent of intestinal

metabolism. For 1,25(OH)2D3, however, the extent of first-pass intestinal removal is small because the total plasma clearance is very low (Table 7-6). The composite data support the SFM as the preferred model over the TM, inferring that there is intestinal route-dependent metabolism, namely, a drug given systemically will be less extracted by the intestine due to the low blood flow rate perfusing the enterocyte region (Cong et al., 2000; Pang, 2003; Pang and Chow, 2012).

The PBPK(SFM)-PD model was extended to the liver and was able to predict changes in Cyp7a1 mRNA or protein expression, as well as cholesterol lowering (Figs. 7-10 and 7-11). The lack of

166

temporal data describing cholesterol levels limited our ability to verify the model output. However, it appears that Cyp7a1 data could describe the reduction in cholesterol upon repeated dosing with

1,25(OH)2D3, with maximum effects (36% reduction) occurring after 8 days. In a population PD analysis of LDL-C levels following treatment with statins, a gradual reduction in LDL-C was also observed, with maximum effects (30-50% reduction) occurring after 30 days (Kakara et al., 2014). Presumably, the incorporation of additional data, including Vdr and Shp expression, would improve model performance. Because hepatic Vdr and Cyp7a1 mRNA expression are upregulated

in a similar temporal fashion following 1,25(OH)2D3 treatment, this upregulation of Vdr could potentially explain the high expression of Cyp7a1 upon repeated dosing. However, an attempt to incorporate this additional data was unsuccessful as the model equations became too complicated

to solve. Nonetheless, these models provide a foundation for predicting 1,25(OH)2D3 disposition for inter-species scaling and for exploration of alternative dosing schemes and routes of administration to describe the dynamics of 1,25(OH)2D3 in its new therapeutic roles.

In conclusion, several biologically plausible models have been described for the quantitative characterization of the roles of Cyp24a1 and Cyp27b1 in regulating the complex PK and

disposition of 1,25(OH)2D3 in mice, although there has been other evidence for altered

1,25(OH)2D3 efficacy due to single nucleotide polymorphisms in the Cyp24a1 gene (Chen et al., 2011; Ramnath et al., 2013). The PBPK(SFM)-PD model has provided a mechanism-based

framework for discerning the tissue-specific disposition characteristics of 1,25(OH)2D3 where the

dynamic effects of 1,25(OH)2D3 are tightly regulated by the endogenous tissue concentrations, providing various platforms to integrate the absorption, distribution, and metabolism/excretion of

1,25(OH)2D3 to biologic effects observed in preclinical studies. This present investigation on the PKPD relationships with respect to renal and intestinal handling of synthesis and metabolism via downregulation of Cyp27b1 and induction of Cyp24a1 has been expanded to view changes in Cyp7a1 and cholesterol lowering, and could be further extended to include other VDR target genes and pharmacological responses.

167

7.6 Appendix

APPENDIX A:

For the simple two-compartment model (Fig. 7-1A): dC V1 = R-kCV+kA-kCV (A1) 1syn12112121011dt

dA2 = k12 C 1 V 1 - k 21 A 2 (A2) dt

Note: k12 for the 60 pmol dose is 1.57x k12 of the 2 pmol dose; k12 for the 120 pmol dose is 5.7x k12 of the 2 pmol dose.

APPENDIX B:

For PKPD model with scaling factors (SM1 and SM2) and Hill coefficients (γ1 and γ2) (Fig. 7-1B):

γ ICγ1 EC2 (1 - max 1 ) (1 +  max 1 ) dC ICγγ11 +C EC γγ22 +C V1 = R50 1 - k C V + k A - k C V 50 1 1syn 12112121011 (B1) dt SM12 SM

APPENDIX C:

For the indirect response model, with scaling factors (SM1 and SM2) to adjust for fold-changes (FC) (Fig. 7-1C):

RFC⋅⋅ kCVFC dC1 syn Cyp27b1 10 1 1 Cyp24a1 V=11211212 -kCV+kA- (C1) dt SM12 SM The rates of change in PD responses (Cyp27b1 or Cyp24a1) can be described as:

dR (C2) = k -k ×R dt in out

where kin is the zero-order rate constant for production of the response and kout is the first-order rate constant for loss of the response. The response variable, R, in this study is the same as the FC of VDR target gene mRNA expression. The inhibition (Equation 1) and induction (Equation 2) functions were included, respectively, to account for the inhibition of Cyp27b1 and induction of

168

Cyp24a1. Inhibition of the Cyp27b1 response variable ( Cyp27b1 ) and induction of the Cyp24a1 kin Cyp24a1 response variable ( k in ) were described as:

dFC I C γ1 Cyp27b1 = kCyp27b1⋅⋅ (1- max p ) - k Cyp27b1 FC (C3) inγγ11 out Cyp27b1 dt IC50 +C p

dFC E C γ2 Cyp24a1 = kCyp24a1⋅⋅ (1+ max p ) - k Cyp24a1 FC (C4) inγγ22 out Cyp24a1 dt EC50 +C p

where Cp is the plasma 1,25(OH)2D3 concentration; kin = kout x baseline, with baseline = 1; Imax and

Emax are the maximal FC for inhibition and induction factors; and IC50 and EC50 are the plasma

concentrations that result in 50% of Imax and Emax. The Emax is the sum of the Emax from the intestine and kidney.

A time-delay function was required to provide a better fit for Cyp27b1 expression, where τ is the

time delay in h and Atransit1 and Atransit2 are the amounts in transit compartments 1 and 2 respectively:

dAtransit FC Cyp27b1 A transit 11 = (C5) dt τ

dAtransit A transit - A transit 212 = (C6) dt τ

APPENDIX D:

The extended PBPK(SFM)-PD model to the following equations incorporate liver Cyp7a1 and cholesterol (Fig. 7-3):

dFC E C γ1 Cyp7a1 = kCyp7a1⋅⋅= (1+ max p ) - k Cyp7a1 FC ; FC 1 (D1) inγγ11 out Cyp7a1 Cyp7a1,baseline dt EC50 +C p

where Emax is the maximum inductive FC, EC50 is the tissue concentration that results in 50% of

Cyp7a1 Emax, CL is the 1,25(OH)2D3 concentration in liver, kin is the zero-order production rate constant

169

Cyp7a1 γ of Cyp7a1, kout is the first-order degradation rate constant of Cyp7a1, and 1 is the Hill coefficient. Initial estimates were obtained from plotting FCCyp7a1 vs. liver 1,25(OH)2D3 (Supplementary Fig. S7-1).

We further employed time-delay functions to provide an improved fit for Cyp7a1 expression, where τ is the time delay in hours and Atransit1 and Atransit2 are the amounts in transit compartments 1 and 2, respectively:

dAtransit FC Cyp7a1 - A transit dAtransit A transit - A transit 11 = ; 212= (D2) dt τ dt τ

Subsequently, Cyp7a1 expression induced degradation of liver cholesterol:

dCholesterol A Cholesterol %reduced − transit2 %reduced = kin,cholesterol k out,cholesterol (D3) dt SM1

where kin,cholesterol is the zero-order production rate constant of cholesterol, kout,cholesterol is the first- order degradation rate constant of cholesterol, and SM1 is a scaling factor.

To account for diurnal variation of Cyp7a1 due to circadian rhythm, we further modified the PBPK(SFM)-PD model using simple cosine functions:

Kk⋅ 2  2π 2π 2π  K =k⋅⋅⋅⋅ IC(2)-amp out,Cyp7a1 cos( t )- sin( t ) (D4) mean out,Cyp7a122 peak⋅ peak  k+(2out,Cyp7a1 π/24)  24 24 kout,Cyp7a1 24 

t-t k=K+Kcos(2⋅⋅peak π) (D5) in,Cyp7a1 mean amp 24 where kin,CYP7A1 is the zero-order production rate constant of Cyp7a1, kout,Cyp7a1 is the first-order degradation rate constant of Cyp7a1, Kmean is the mean input rate, IC(2) is the baseline value of the FC response, Kamp is the amplitude of input rate, t is the decimal clock time, and tpeak is the peak time. These parameters were used to modify Equation D1. Initial estimates to describe Cyp7a1 mRNA and protein circadian rhythm, including mean input rate, amplitude, and peak time, were obtained from previously published results (Chow et al., 2014). However, these basic

170

equations could not adequately predict the diurnal variation (data was not shown), which may require further modification with core clock genes, circadian rhythms of the cholesterol synthesis enzyme, and consideration of time of feeding and types of diets.

7.7 Acknowledgments

This work was supported by the Canadian Institutes of Health Research (KSP), the National Sciences and Engineering Research Council of Canada (HPQ, ECYC), and the Ontario Graduate Scholarship Program (HPQ, QJY). We thank Dr. Matthew R. Durk for assistance in the study.

7.8 Supplementary material

Supplementary Table S7-1. Fitted parameters [estimate (CV%)] obtained by modifying the liver compartment of the PBPK(SFM)-PD model to include Cyp7a1 mRNA or Cyp7a1 protein (Fig. 7-3)

Cyp7a1 Cyp7a1 Parameter Definition mRNA protein

kin,Cyp7a1 or a Turnover rate constants of hepatic Cyp7a1 0.04 (54) 0.08 (88) kout,Cyp7a1 E max,Cyp7a1 E for Cyp7a1 fold-change vs. liver 1,25(OH) D 148 (42) 13.5 (28) (fold-change) max 2 3 EC 50,Cyp7a1 EC for Cyp7a1 fold-change vs. liver 1,25(OH) D 307 (35) 771 (26) (pmol·kg-1 tissue) 50 2 3

γ1 Hill coefficient for indirect response function of Cyp7a1 0.81 (58) 0.54 (31) kin,cholesterol or a Turnover rate constant of liver cholesterol 0.04 (32) 0.02 (14) kout,cholesterol SM1 Scaling factor for liver cholesterol turnover 9.8 (31) 1.3 (23) τ (h) Time delay function 0.35 (131) 0.18 (39) AIC Akaike information criterion 736.5 405.4 WSSR Weighted sum of squared residuals 84.5 85.3 df Degrees of freedom 23 21 F-test value Calculated F score -0.10 Note: the full PBPK(SFM)-PD model (Fig. 7-2) was modified and the fitted parameters obtained for all other compartments (Table 7-2) were fixed in this extended model. a Normally the units for the zero-order production (kin) and first-order degradation (kout) rate constants, are mass per unit time and 1/time, respectively. However, when the value is normalized to the baseline value, the rate constants become unitless.

171

Supplementary Figure S7-1. FC in hepatic Cyp7a1 (A) mRNA and (B) protein expression versus liver 1,25(OH)2D3 concentrations following all four doses of 120 pmol 1,25(OH)2D3 i.p. [data of (Chow et al., 2014)]. Equation D1 for induction was used to describe FCCyp7a1; baseline values of FCCyp7a1 = 1. Initial estimates for Cyp7a1 mRNA: Emax,Cyp7a1 = 14-fold, EC50,Cyp7a1 = 400 pmol/kg; and Cyp7a1 protein: Emax,Cyp7a1 = 6-fold, EC50,Cyp7a1 = 1342 pmol/kg.

7.9 Statement of significance of Chapter 7

The therapeutic use of 1,25(OH)2D3 is often limited by the propensity of 1,25(OH)2D3 to cause hypercalcemia and adverse effects. Recently, our laboratory identified a role of the 1,25(OH)2D3- activated VDR in lowering cholesterol (Chow et al., 2014). Thus, an understanding of the PK and

PD of exogenously administered 1,25(OH)2D3 is critical for the prediction of a proper dose and dosing regimen for potential therapeutic uses. Information on the PK of 1,25(OH)2D3 is equivocal.

Some of the discrepancies in the reported PK parameters of 1,25(OH)2D3 may exist due to dose- dependent PK and altered PD with dose and route.

In this chapter, we first identified that the kinetics and disposition of exogenously administered

1,25(OH)2D3 are dose-dependent, a consequence of its ability to regulate its own levels through induction of the degradation enzyme (Cyp24a1) and inhibition of the synthesis enzyme (Cyp27b1) with increasing dose. Here, we showed that a PKPD model that incorporated the dose-dependent effect of Cyp24a1 on clearance (or k10) and Cyp27b1 on Rsyn was superior in predicting observed

172

data when compared with the simple two-compartment model. Next, we developed a PBPK-PD model to represent 1,25(OH)2D3 disposition in a physiologic system. Compared with the PKPD model, the PBPK-PD model was more biologically plausible as it utilized blood flow and estimated parameters using relevant tissue 1,25(OH)2D3 concentrations rather than plasma 1,25(OH)2D3. Additionally, our data proved that the PBPK(SFM)-PD model was superior compared with the PBPK(TM)-PD model in describing route-dependent intestinal metabolism/disposition of

1,25(OH)2D3. An extension of the PBPK(SFM)-PD model was developed and able to adequately describe liver Cyp7a1 and cholesterol levels following repeated treatment with 1,25(OH)2D3.

Therefore, the PBPK(SFM)-PD model may be used to predict 1,25(OH)2D3 disposition for exploration of alternative dosing regimens and routes of administration to describe 1,25(OH)2D3 and its novel therapeutic applications.

My contributions to this chapter include design of experiments, maintenance and treatment of mice, tissue harvesting, extractions and determination of liver cholesterol and 1,25(OH)2D3 in plasma and various tissues, mRNA analysis in both i.p. and i.v. studies, model fitting for the two- compartment and extended PBPK(SFM)-PD models, and writing the paper. Qi J. Yang was responsible for model fitting for the PKPD, indirect response, PBPK(TM)-PD, and PBPK(SFM)- PD models and writing the paper. Edwin C.Y. Chow contributed to the design of experiments, maintenance and treatment of mice, tissue harvesting, mRNA analysis, and 1,25(OH)2D3 measurements of i.p. data. Stacie Y. Hoi performed mRNA analysis in the i.v. study. Vidya Ramakrishnan and Yanguang Cao contributed to model fitting of the minimal PBPK(TM)-PD and minimal PBPK(SFM)-PD models (Appendix VI; data not shown in this chapter) and writing the paper. Donald E. Mager contributed to analyses of data and writing the paper. K. Sandy Pang contributed to design of experiments, analyses of data, and writing the paper.

173

Chapter 8

8 Discussion and Conclusions

The VDR plays a critical role in human health and disease. In addition to its traditional role in the regulation of mineral homeostasis, the VDR has been identified as a key regulator of nuclear receptors, transporters, and enzymes that are involved in the disposition of both xenobiotic and endogenous compounds. The VDR is further demonstrated to play important roles in the prevention of cancer, cardiovascular disease, inflammation, and diabetes (Valdivielso et al., 2009).

More recently, the 1,25(OH)2D3-liganded VDR is shown to increase efflux of amyloid-β to improve Alzheimer’s disease (Durk et al., 2014) and lowers cholesterol levels (Chow et al., 2014). As such, it is not surprising that vitamin D deficiency is associated with a number of diseases. What is surprising, however, is the fact that vitamin D deficiency remains highly prevalent around the world even though it is so easily diagnosed and treated. Furthermore, although there are many clinical studies that reported on associations between vitamin D deficiency and disease, there are few clinical studies that have identified causation and even fewer preclinical studies that have identified plausible biological mechanisms to explain the pathogenesis of disease. Clearly, further investigation is required with respect to the role of vitamin D status and the VDR on mechanisms of disease.

In Chapter 3, we examined the direct effects of the VDR in the mouse kidney, intestine, and bone.

Here, we first demonstrated that exogenously administered 1,25(OH)2D3 could rapidly equilibrate between plasma and tissue (Fig. 3-1). Therefore, the resultant temporal profiles of VDR target genes that paralleled the rise-and-fall of 1,25(OH)2D3 levels was attributed to activation of the

VDR by 1,25(OH)2D3. Interestingly, we observed that repeated doses of 1,25(OH)2D3 increased its own elimination by induction of Cyp24a1 in tissues, as did others, although none have explored the mechanism nor how that alters the PK of 1,25(OH)2D3. We took on this challenge and further investigated the mechanism and consequences (see Chapter 7). The ability to determine exact

1,25(OH)2D3 levels in different tissues is crucial for the elucidation of the roles of 1,25(OH)2D3- activated VDR in the body, including cholesterol-regulating genes (Shp and Cyp7a1) for cholesterol homeostasis under normal (Chapter 4) and vitamin D-deficient (Chapter 5) conditions.

174

Overall, we showed that VDR-related gene changes were directly correlated with tissue (liver) levels of 1,25(OH)2D3. In Chapter 4, we demonstrated that exogenously administered 1,25(OH)2D3 could equilibrate rapidly between plasma and liver to activate the VDR to directly inhibit Shp and increase Cyp7a1 and cholesterol metabolism (Figs. 4-1 and 4-2). The experiments with Fxr(-/-) mice were crucial to assert that the mechanism was independent of the Fxr. Hence, we identified a direct role of the VDR in cholesterol lowering and uncovered a novel therapeutic target for hypercholesterolemia.

We revisited this finding in Chapter 5 and tested the corollary of whether vitamin D deficiency and the associated reduced VDR expression could cause hypercholesterolemia. Here, we found that vitamin D-deficient mice, as defined by low 25(OH)D3 and 1,25(OH)2D3 levels, had significantly elevated plasma and liver cholesterol levels (Fig. 5-2). These mice also displayed downregulated Vdr and Cyp7a1 and elevated Shp expression (Fig. 5-3). A summary of changes is depicted in Fig. 8-1. When we plotted Vdr, Cyp7a1, Shp, and cholesterol vs. liver 1,25(OH)2D3, we observed direct positive (Vdr and Cyp7a1) and inverse (Shp and cholesterol) correlations (Fig.

5-4). Meanwhile, significant inverse correlations between 1,25(OH)2D3 and Cyp7a1 vs. cholesterol levels were also identified in the mouse liver (Fig. 5-5). Interestingly, similar correlations were obtained in human liver tissues, suggesting that a similar mechanism of vitamin D deficiency-induced hypercholesterolemia could exist in humans and could be used to explain the commonly reported association between vitamin D deficiency and hypercholesterolemia in clinical studies. As such, it appears that the vitamin D status directly affects cholesterol levels through alterations in the Vdr-Shp-Cyp7a1 cascade. We further tested this hypothesis by intervention of vitamin D-deficient mice with vitamin D3 and 1,25(OH)2D3 to replenish vitamin D status back to normal levels (Figs. 5-7 and 5-8). Following treatment, the altered expression of cholesterol-regulating genes and elevated cholesterol levels that were observed in vitamin D- deficient mice was reversed. Moreover, we observed a similar effect in mice that were fed a high fat/high cholesterol diet. Here, cholesterol levels were elevated compared with normal diet controls and vitamin D-deficient diets further exacerbated this effect, indicating that our mechanism of vitamin D deficiency-induced hypercholesterolemia persists in both normal and hypercholesterolemic models. Overall, we have identified the VDR as a novel therapeutic target for hypercholesterolemia, since the administration of vitamin D3 and 1,25(OH)2D3 was able to inhibit Shp and increase Cyp7a1 expression to reduce elevated cholesterol levels back to baseline.

175

Figure 8-1. Summary of changes in cholesterol-regulating genes, cholesterol, and bile acid concentrations under vitamin D-deficient conditions.

The therapeutic utility of 1,25(OH)2D3, however, is severely restricted by its tendency to induce toxicity in the form of hypercalcemia. In Chapter 6, we investigated the efficacy of vitamin D analogs on the expression of cholesterol-regulating genes. While initial luciferase reporter assay results suggested a similar potency between 25(OH)D3 and 1α(OH)D3 (Fig. 6-1A), we found that

1α(OH)D3 was much more effective in inducing VDR genes in Caco-2 cells and increasing Cyp7a1 for cholesterol lowering in mice in vivo. It turns out that the presence of bioactivation enzymes towards activation of these vitamin D analogs to 1,25(OH)2D3 overstated the potencies. The conversion of 25(OH)D3 to 1,25(OH)2D3 is mediated by CYP27B1, while conversion of 1α(OH)D3 is mediated by CYP2R1. In vitro, when we compared basal expressions of CYP27B1 and CYP2R1 in different cell lines, we found a discrepancy that could explain previous results. HEK293 cells displayed significantly higher expression of CYP27B1 than Caco-2 cells, while Caco-2 cells had higher levels of CYP2R1 (Fig. 6-3). Therefore, the initial finding from the luciferase reporter assay that these analogs would behave similarly in vivo was influenced by the high basal levels of

CYP27B1 in HEK293 cells that may have exaggerated the potency of 25(OH)D3 vs. 1α(OH)D3.

176

As such, we recommend a full characterization of the bioactivation enzymes in different cell systems prior to any interpretation of data following treatment with vitamin D analogs, as potency is dependent upon bioactivation to 1,25(OH)2D3. In contrast to direct administration with

1,25(OH)2D3 that would immediately trigger its own degradation, treatment with 1α(OH)D3 may well sustain 1,25(OH)2D3 pools in the body via continuous bioactivation. In mice in vivo, we directly measured 1,25(OH)2D3 levels formed following treatment with these vitamin D analogs in both plasma and liver and found that 1α(OH)D3 administration resulted in significantly higher

1,25(OH)2D3 levels in both plasma and liver, while 25(OH)D3 administration only increased

1,25(OH)2D3 in the liver (Fig. 6-4). The resultant levels of 1,25(OH)2D3 were thought to be responsible for the observed cholesterol lowering effects. A significant inverse correlation between levels of cholesterol and 1,25(OH)2D3 in the liver was also observed (Fig. 6-5), a finding that supports the notion that levels of 1,25(OH)2D3 are more directly related to cholesterol in the liver.

Another method to circumvent hypercalcemia was to alter the dosing regimen or route of administration of 1,25(OH)2D3 by understanding the PK and PD of 1,25(OH)2D3. In Chapter 7, we first developed a PKPD model to describe the dose-dependent effects of intravenously administered 1,25(OH)2D3 on the synthesis and degradation enzymes. Here, mice were treated i.v. with different doses of 1,25(OH)2D3 and a PD-linked model was developed to best describe the altered disposition of 1,25(OH)2D3 with increasing doses by incorporating its dose-dependent actions on the inhibition of Cyp27b1 (synthesis) and induction of Cyp24a1 (degradation). Using the rich i.p. data set from Chapter 3, our laboratory further developed a PBPK-PD model that best describes multi-tissue (kidney, intestine, liver, and brain) synthesis and degradation of

1,25(OH)2D3. In this model, we compared the model performance of a traditional vs. segregated flow model (SFM) of the intestine, which further provided insight on the impact of route of administration. We showed that the PBPK(SFM)-PD model was superior in predicting the route- dependent intestinal removal/distribution of 1,25(OH)2D3 through simulations of i.v. data. Overall, we were able to utilize and extend this biologically plausible model to the liver to describe Cyp7a1 effects on cholesterol lowering following repeated i.p. administration of 1,25(OH)2D3.

In conclusion, we have demonstrated that VDR-related gene changes are directly correlated with tissue 1,25(OH)2D3 levels. Furthermore, we have identified a direct role of the VDR in cholesterol lowering via inhibition of Shp and induction of Cyp7a1, a mechanism that is independent of the

177

FXR. In the mouse liver, we found that 1,25(OH)2D3 levels were directly correlated with Vdr, Cyp7a1, Shp, and cholesterol levels. Interestingly, liver cholesterol levels were inversely correlated with 1,25(OH)2D3 and Cyp7a1/CYP7A1 in both mouse and human liver tissues. These composite findings have provided valuable insight into the widely noted association observed clinically between vitamin D deficiency and hypercholesterolemia. Although we have identified the VDR to be a therapeutic target for hypercholesterolemia, we acknowledge the fact that treatment with 1,25(OH)2D3 is limited by its propensity to induce hypercalcemia. Therefore, the efficacy of vitamin D analogs that could potentially elicit desirable therapeutic effects without hypercalcemia were explored. Here, it was determined that potency of the precursors of

1,25(OH)2D3 was highly dependent upon the expression of the bioactivation enzymes that existed in the different systems. Future studies could use alternate VDR ligands that do not rely on bioactivation to 1,25(OH)2D3 for efficacy. Alternatively, the development of PKPD and PBPK-

PD models to predict the disposition of exogenously administered 1,25(OH)2D3 provided insight into strategies to avoid hypercalcemia by altering 1,25(OH)2D3 dosing schedules and/or routes of administration. These models could also be used to describe and predict the dynamics of

1,25(OH)2D3 in its newly identified therapeutic role in cholesterol lowering by induction of Cyp7a1. Future modifications of the PBPK-PD model could incorporate other pharmacological effects, including 1,25(OH)2D3-mediated efflux of amyloid-β in the prevention of Alzheimer’s disease. Overall, our findings have demonstrated the effects of vitamin D deficiency on hypercholesterolemia, highlighting the importance of maintaining sufficient vitamin D levels. Moreover, we have established the VDR as a potential therapeutic target for the treatment of hypercholesterolemia. Future studies could further investigate and optimize a dosing regimen for a combination therapy with 1,25(OH)2D3 or vitamin D3 and atorvastatin for synergistic cholesterol lowering effects. For additional mechanistic insights, future studies on vitamin D deficiency could utilize LDL receptor knockout mice. Furthermore, total cholesterol levels could be further dissected into different lipoprotein fractions (LDL vs. HDL). Overall, the relationship between vitamin D deficiency and hypercholesterolemia could be correlated to clinical situations to demonstrate the importance of the VDR in hypercholesterolemia and how vitamin D deficiency- associated hypercholesterolemia could be averted.

178

References

Abe J, Nakano T, Nishii Y, Matsumoto T, Ogata E, and Ikeda K (1991) A novel vitamin D3 analog, 22-oxa-1,25-dihydroxyvitamin D3, inhibits the growth of human breast cancer in vitro and in vivo without causing hypercalcemia. Endocrinology 129:832-837.

Ahmed S, Johnson CS, Rueger RM, and Trump DL (2002) Calcitriol (1,25-dihydroxycholecalciferol) potentiates activity of mitoxantrone/dexamethasone in an androgen independent prostate cancer model. J Urol 168:756-761.

Akeno N, Saikatsu S, Kimura S, and Horiuchi N (1994) Induction of vitamin D 24-hydroxylase messenger RNA and activity by 22-oxacalcitriol in mouse kidney and duodenum. Possible role in decrease of plasma 1α,25-dihydroxyvitamin D3. Biochem Pharmacol 48:2081-2090.

Al-Ajlan A, Krishnaswamy S, Alokail MS, Aljohani NJ, Al-Serehi A, Sheshah E, Alshingetti NM, Fouda M, Turkistani IZ, and Al-Daghri NM (2015) Vitamin D deficiency and dyslipidemia in early pregnancy. BMC Pregnancy Childbirth 15:314.

Albert DM, Scheef EA, Wang S, Mehraein F, Darjatmoko SR, Sorenson CM, and Sheibani N (2007) Calcitriol is a potent inhibitor of retinal neovascularization. Invest Ophthalmol Vis Sci 48:2327-2334.

Alnouti Y, Csanaky IL, and Klaassen CD (2008) Quantitative-profiling of bile acids and their conjugates in mouse liver, bile, plasma, and urine using LC-MS/MS. J Chromatogr B Analyt Technol Biomed Life Sci 873:209-217.

Alphonse PA and Jones PJ (2016) Revisiting Human Cholesterol Synthesis and Absorption: The Reciprocity Paradigm and its Key Regulators. Lipids 51:519-536.

Altmann SW, Davis HR, Jr., Zhu LJ, Yao X, Hoos LM, Tetzloff G, Iyer SP, Maguire M, Golovko A, Zeng M, Wang L, Murgolo N, and Graziano MP (2004) Niemann-Pick C1 Like 1 protein is critical for intestinal cholesterol absorption. Science 303:1201-1204.

Andress DL (2006) Vitamin D in chronic kidney disease: a systemic role for selective vitamin D receptor activation. Kidney Int 69:33-43.

Armbrecht HJ and Boltz MA (1991) Expression of 25-hydroxyvitamin D 24-hydroxylase cytochrome P450 in kidney and intestine. Effect of 1,25-dihydroxyvitamin D and age. FEBS Lett 292:17- 20.

Asemi Z, Hashemi T, Karamali M, Samimi M, and Esmaillzadeh A (2013) Effects of vitamin D supplementation on glucose metabolism, lipid concentrations, inflammation, and oxidative stress in gestational diabetes: a double-blind randomized controlled clinical trial. Am J Clin Nutr 98:1425-1432.

Bell NH (1998) Renal and nonrenal 25-hydroxyvitamin D-1alpha-hydroxylases and their clinical significance. J Bone Miner Res 13:350-353.

179

Bergman T and Postlind H (1990) Characterization of pig kidney microsomal cytochrome P-450 catalysing 25-hydroxylation of vitamin D3 and C27 steroids. Biochem J 270:345-350.

Berry D and Hypponen E (2011) Determinants of vitamin D status: focus on genetic variations. Curr Opin Nephrol Hypertens 20:331-336.

Bettoun DJ, Burris TP, Houck KA, Buck DW, 2nd, Stayrook KR, Khalifa B, Lu J, Chin WW, and Nagpal S (2003) Retinoid X receptor is a nonsilent major contributor to vitamin D receptor- mediated transcriptional activation. Mol Endocrinol 17:2320-2328.

Bi X, Tey SL, Leong C, Quek R, and Henry CJ (2016) Prevalence of Vitamin D Deficiency in Singapore: Its Implications to Cardiovascular Risk Factors. PLoS One 11:e0147616.

Bianco SD, Peng JB, Takanaga H, Suzuki Y, Crescenzi A, Kos CH, Zhuang L, Freeman MR, Gouveia CH, Wu J, Luo H, Mauro T, Brown EM, and Hediger MA (2007) Marked disturbance of calcium homeostasis in mice with targeted disruption of the Trpv6 calcium channel gene. J Bone Miner Res 22:274-285.

Biggerstaff KD and Wooten JS (2004) Understanding lipoproteins as transporters of cholesterol and other lipids. Adv Physiol Educ 28:105-106.

Bischoff-Ferrari HA, Dawson-Hughes B, Stocklin E, Sidelnikov E, Willett WC, Edel JO, Stahelin HB, Wolfram S, Jetter A, Schwager J, Henschkowski J, von Eckardstein A, and Egli A (2012) Oral supplementation with 25(OH)D3 versus vitamin D3: effects on 25(OH)D levels, lower extremity function, blood pressure, and markers of innate immunity. J Bone Miner Res 27:160-169.

Bligh EG and Dyer WJ (1959) A rapid method of total lipid extraction and purification. Can J Biochem Physiol 37:911-917.

Bonakdaran S, Nejad AF, Abdol-Reza V, Hatefi A, and Shakeri M (2013) Impact of oral 1,25- dihydroxy vitamin D (calcitriol) replacement therapy on coronary artery risk factors in type 2 diabetic patients. Endocr Metab Immune Disord Drug Targets 13:295-300.

Bonventre JV (2003) Dedifferentiation and proliferation of surviving epithelial cells in acute renal failure. J Am Soc Nephrol 14 Suppl 1:S55-61.

Bouillon R, Carmeliet G, Verlinden L, van Etten E, Verstuyf A, Luderer HF, Lieben L, Mathieu C, and Demay M (2008) Vitamin D and human health: lessons from vitamin D receptor null mice. Endocr Rev 29:726-776.

Bouillon R, Okamura WH, and Norman AW (1995) Structure-function relationships in the vitamin D endocrine system. Endocr Rev 16:200-257.

Bousquet-Melou A, Galitzky J, Lafontan M, Berlan M (1995) Control of lipolysis in intra-abdominal fat cells of nonhuman primates: comparison with humans. J Lipid Res 36:451-61.

180

Boxenbaum HG, Riegelman S, and Elashoff RM (1974) Statistical estimations in pharmacokinetics. J Pharmacokinet Biopharm 2:123-148.

Brandi L, Egfjord M, and Olgaard K (2002) Pharmacokinetics of 1,25(OH)2D3 and 1α(OH)D3 in normal and uraemic men. Nephrol Dial Transplant 17:829-842.

Brenza HL and DeLuca HF (2000) Regulation of 25-hydroxyvitamin D3 1α-hydroxylase gene expression by parathyroid hormone and 1,25-dihydroxyvitamin D3. Arch Biochem Biophys 381:143-152.

Brown AJ and Slatopolsky E (2008) Vitamin D analogs: therapeutic applications and mechanisms for selectivity. Mol Aspects Med 29:433-452.

Brown RP, Delp MD, Lindstedt SL, Rhomberg LR, and Beliles RP (1997) Physiological parameter values for physiologically based pharmacokinetic models. Toxicol Ind Health 13:407-484.

Carbone LD, Rosenberg EW, Tolley EA, Holick MF, Hughes TA, Watsky MA, Barrow KD, Chen TC, Wilkin NK, Bhattacharya SK, Dowdy JC, Sayre RM, and Weber KT (2008) 25- Hydroxyvitamin D, cholesterol, and ultraviolet irradiation. Metabolism 57:741-748.

Carrillo-Lopez N, Alvarez-Hernandez D, Gonzalez-Suarez I, Roman-Garcia P, Valdivielso JM, Fernandez-Martin JL, and Cannata-Andia JB (2008) Simultaneous changes in the calcium- sensing receptor and the vitamin D receptor under the influence of calcium and calcitriol. Nephrol Dial Transplant 23:3479-3484.

Cella LK, Van Cauter E, and Schoeller DA (1995) Diurnal rhythmicity of human cholesterol synthesis: normal pattern and adaptation to simulated “jet lag”. Am J Physiol 269:E489-498.

Chan JC and DeLuca HR (1979) Calcium and parathyroid disorders in children. Chronic renal failure and treatment with calcitriol. JAMA 241:1242-1244.

Chen F, Ma L, Dawson PA, Sinal CJ, Sehayek E, Gonzalez FJ, Breslow J, Ananthanarayanan M, Shneider BL (2003) Liver receptor homologue-1 mediates species- and cell line-specific bile acid-dependent negative feedback regulation of the apical sodium-dependent bile acid transporter. J Biol Chem 278:19909-19916.

Chen G, Kim SH, King AN, Zhao L, Simpson RU, Christensen PJ, Wang Z, Thomas DG, Giordano TJ, Lin L, Brenner DE, Beer DG, and Ramnath N (2011) CYP24A1 is an independent prognostic marker of survival in patients with lung adenocarcinoma. Clin Cancer Res 17:817- 826.

Chen KS and DeLuca HF (1995) Cloning of the human 1α,25-dihydroxyvitamin D3 24-hydroxylase gene promoter and identification of two vitamin D-responsive elements. Biochim Biophys Acta 1263:1-9.

181

Chen X, Chen F, Liu S, Glaeser H, Dawson PA, Hofmann AF, Kim RB, Shneider BL, and Pang KS (2006) Transactivation of rat apical sodium-dependent bile acid transporter and increased bile acid transport by 1α,25-dihydroxyvitamin D3 via the vitamin D receptor. Mol Pharmacol 69:1913-1923.

Cheng JB, Motola DL, Mangelsdorf DJ, and Russell DW (2003) De-orphanization of cytochrome P450 2R1: a microsomal vitamin D 25-hydroxylase. J Biol Chem 278:38084-38093.

Chiang JY (2002) Bile acid regulation of gene expression: roles of nuclear hormone receptors. Endocr Rev 23:443-463.

Chiang JY (2003) Bile acid regulation of hepatic physiology: III. Bile acids and nuclear receptors. Am J Physiol Gastrointest Liver Physiol 284:G349-356.

Chiang JY (2004) Regulation of bile acid synthesis: pathways, nuclear receptors, and mechanisms. J Hepatol 40:539-551.

Chiang JY, Kimmel R, and Stroup D (2001) Regulation of cholesterol 7α-hydroxylase gene (CYP7A1) transcription by the liver orphan receptor (LXRα). Gene 262:257-265.

Chiang JY and Stroup D (1994) Identification and characterization of a putative bile acid-responsive element in cholesterol 7α-hydroxylase gene promoter. J Biol Chem 269:17502-17507.

Chow EC, Durk MR, Cummins CL, and Pang KS (2011a) 1α,25-dihydroxyvitamin D3 up-regulates P-glycoprotein via the vitamin D receptor and not farnesoid X receptor in both Fxr(-/-) and fxr(+/+) mice and increased renal and brain efflux of digoxin in mice in vivo. J Pharmacol Exp Ther 337:846-859.

Chow EC, Durk MR, Maeng HJ, and Pang KS (2013a) Comparative effects of 1α-hydroxyvitamin D3 and 1,25-dihydroxyvitamin D3 on transporters and enzymes in Fxr(+/+) and Fxr(-/-) mice. Biopharm Drug Dispos 34:402-416.

Chow EC, Maeng HJ, Liu S, Khan AA, Groothuis GM, and Pang KS (2009) 1α,25-Dihydroxyvitamin D3 triggered vitamin D receptor and farnesoid X receptor-like effects in rat intestine and liver in vivo. Biopharm Drug Dispos 30:457-475.

Chow EC, Magomedova L, Quach HP, Patel R, Durk MR, Fan J, Maeng HJ, Irondi K, Anakk S, Moore DD, Cummins CL, and Pang KS (2014) Vitamin D receptor activation down-regulates the small heterodimer partner and increases CYP7A1 to lower cholesterol. Gastroenterology 146:1048-1059.

Chow EC, Quach HP, Vieth R, and Pang KS (2013b) Temporal changes in tissue 1α,25- dihydroxyvitamin D3, vitamin D receptor target genes, and calcium and PTH levels after 1,25(OH)2D3 treatment in mice. Am J Physiol Endocrinol Metab 304:E977-989.

Chow EC, Sondervan M, Jin C, Groothuis GM, and Pang KS (2011b) Comparative effects of doxercalciferol (1α-hydroxyvitamin D2) versus calcitriol (1α,25-dihydroxyvitamin D3) on the expression of transporters and enzymes in the rat in vivo. J Pharm Sci 100:1594-1604.

182

Chow EC, Sun H, Khan AA, Groothuis GM, and Pang KS (2010) Effects of 1α,25-dihydroxyvitamin D3 on transporters and enzymes of the rat intestine and kidney in vivo. Biopharm Drug Dispos 31:91-108.

Chun RF (2012) New perspectives on the vitamin D binding protein. Cell Biochem Funct 30:445- 456.

Clements MR, Davies M, Hayes ME, Hickey CD, Lumb GA, Mawer EB, and Adams PH (1992) The role of 1,25-dihydroxyvitamin D in the mechanism of acquired vitamin D deficiency. Clin Endocrinol (Oxf) 37:17-27.

Colston KW, Chander SK, Mackay AG, and Coombes RC (1992a) Effects of synthetic vitamin D analogues on breast cancer cell proliferation in vivo and in vitro. Biochem Pharmacol 44:693- 702.

Colston KW, Mackay AG, James SY, Binderup L, Chander S, and Coombes RC (1992b) EB1089: a new vitamin D analogue that inhibits the growth of breast cancer cells in vivo and in vitro. Biochem Pharmacol 44:2273-2280.

Cong D, Doherty M, and Pang KS (2000) A new physiologically based, segregated-flow model to explain route-dependent intestinal metabolism. Drug Metab Dispos 28:224-235.

Crestani M, Sadeghpour A, Stroup D, Galli G, and Chiang JY (1998) Transcriptional activation of the cholesterol 7α-hydroxylase gene (CYP7A) by nuclear hormone receptors. J Lipid Res 39:2192-2200.

Cui M, Li Q, Johnson R, and Fleet JC (2012) Villin promoter-mediated transgenic expression of transient receptor potential cation channel, subfamily V, member 6 (TRPV6) increases intestinal calcium absorption in wild-type and vitamin D receptor knockout mice. J Bone Miner Res 27:2097-2107.

Cutillas-Marco E, Prosper AF, Grant WB, and Morales-Suarez-Varela MM (2013) Vitamin D status and hypercholesterolemia in Spanish general population. Dermatoendocrinol 5:358-362.

Davies B and Morris T (1993) Physiological parameters in laboratory animals and humans. Pharm Res 10:1093-1095.

Davis W, Rockway S, and Kwasny M (2009) Effect of a combined therapeutic approach of intensive lipid management, omega-3 fatty acid supplementation, and increased serum 25 (OH) vitamin D on coronary calcium scores in asymptomatic adults. Am J Ther 16:326-332.

Dayneka NL, Garg V, and Jusko WJ (1993) Comparison of four basic models of indirect pharmacodynamic responses. J Pharmacokinet Biopharm 21:457-478.

Deeb KK, Trump DL, and Johnson CS (2007) Vitamin D signalling pathways in cancer: potential for anticancer therapeutics. Nat Rev Cancer 7:684-700.

183

DeLuca HF (1988) The vitamin D story: a collaborative effort of basic science and clinical medicine. FASEB J 2:224-236.

DeLuca HF, Prahl JM, and Plum LA (2010) 1,25-Dihydroxyvitamin D is not responsible for toxicity caused by vitamin D or 25-hydroxyvitamin D. Arch Biochem Biophys 505:226-230. den Dekker E, Hoenderop JG, Nilius B, and Bindels RJ (2003) The epithelial calcium channels, TRPV5 & TRPV6: from identification towards regulation. Cell Calcium 33:497-507.

Ding N, Yu RT, Subramaniam N, Sherman MH, Wilson C, Rao R, Leblanc M, Coulter S, He M, Scott C, Lau SL, Atkins AR, Barish GD, Gunton JE, Liddle C, Downes M, and Evans RM (2013) A vitamin D receptor/SMAD genomic circuit gates hepatic fibrotic response. Cell 153:601- 613.

Dingemanse J and Appel-Dingemanse S (2007) Integrated pharmacokinetics and pharmacodynamics in drug development. Clin Pharmacokinet 46:713-737.

Dobnig H, Pilz S, Scharnagl H, Renner W, Seelhorst U, Wellnitz B, Kinkeldei J, Boehm BO, Weihrauch G, and Maerz W (2008) Independent association of low serum 25-hydroxyvitamin D and 1,25-dihydroxyvitamin D levels with all-cause and cardiovascular mortality. Arch Intern Med 168:1340-1349.

Doherty MM and Pang KS (1997) First-pass effect: significance of the intestine for absorption and metabolism. Drug Chem Toxicol 20:329-344.

Doherty MM and Pang KS (2000) Route-dependent metabolism of morphine in the vascularly perfused rat small intestine preparation. Pharm Res 17:291-298.

Drocourt L, Ourlin JC, Pascussi JM, Maurel P, and Vilarem MJ (2002) Expression of CYP3A4, CYP2B6, and CYP2C9 is regulated by the vitamin D receptor pathway in primary human hepatocytes. J Biol Chem 277:25125-25132.

Dunlop TW, Vaisanen S, Frank C, Molnar F, Sinkkonen L, and Carlberg C (2005) The human peroxisome proliferator-activated receptor delta gene is a primary target of 1α,25- dihydroxyvitamin D3 and its nuclear receptor. J Mol Biol 349:248-260.

Durk MR, Chan GN, Campos C, Peart J, Chow EC, Lee E, Cannon RE, Bendayan R, Miller DS, and Pang KS (2012) 1α,25-Dihydroxyvitamin D3-liganded vitamin D receptor increases expression and transport activity of P-glycoprotein in isolated rat brain capillaries and human and rat brain microvessel endothelial cells. J Neurochem 123:944-953.

Durk MR, Han K, Chow EC, Ahrens R, Henderson JT, Fraser PE, and Pang KS (2014) 1α,25- Dihydroxyvitamin D3 reduces cerebral amyloid-β accumulation and improves cognition in mouse models of Alzheimer's disease. J Neurosci 34:7091-7101.

Dusso AS, Brown AJ, and Slatopolsky E (2005) Vitamin D. Am J Physiol Renal Physiol 289:F8-28.

184

Echchgadda I, Song CS, Roy AK, and Chatterjee B (2004) Dehydroepiandrosterone sulfotransferase is a target for transcriptional induction by the vitamin D receptor. Mol Pharmacol 65:720- 729.

Eisman JA and DeLuca HF (1977) Intestinal 1,25-dihydroxyvitamin D3 binding protein: specificity of binding. Steroids 30:245-257.

Elferink MG, Olinga P, van Leeuwen EM, Bauerschmidt S, Polman J, Schoonen WG, Heisterkamp SH, and Groothuis GM (2011) Gene expression analysis of precision-cut human liver slices indicates stable expression of ADME-Tox related genes. Toxicol Appl Pharmacol 253:57-69.

Eloranta JJ, Hiller C, Juttner M, and Kullak-Ublick GA (2012) The SLCO1A2 gene, encoding human organic anion-transporting polypeptide 1A2, is transactivated by the vitamin D receptor. Mol Pharmacol 82:37-46.

Eloranta JJ, Zair ZM, Hiller C, Hausler S, Stieger B, and Kullak-Ublick GA (2009) Vitamin D3 and its nuclear receptor increase the expression and activity of the human proton-coupled folate transporter. Mol Pharmacol 76:1062-1071.

Engelman CD, Fingerlin TE, Langefeld CD, Hicks PJ, Rich SS, Wagenknecht LE, Bowden DW, and Norris JM (2008) Genetic and environmental determinants of 25-hydroxyvitamin D and 1,25- dihydroxyvitamin D levels in Hispanic and African Americans. J Clin Endocrinol Metab 93:3381-3388.

Fan J, Chen S, Chow EC, and Pang KS (2010) PBPK modeling of intestinal and liver enzymes and transporters in drug absorption and sequential metabolism. Curr Drug Metab 11:743-761.

Fan J, Liu S, Du Y, Morrison J, Shipman R, and Pang KS (2009) Up-regulation of transporters and enzymes by the vitamin D receptor ligands, 1α,25-dihydroxyvitamin D3 and vitamin D analogs, in the Caco-2 cell monolayer. J Pharmacol Exp Ther 330:389-402.

Feldman D, Pike JW, and Glorieux FH (2005) Vitamin D. Elsevier Academic Press, Amsterdam; Boston.

Ferraris RP, Villenas SA, and Diamond J (1992) Regulation of brush-border enzyme activities and enterocyte migration rates in mouse small intestine. Am J Physiol 262:G1047-1059.

Ferrell JM and Chiang JY (2015) Short-term circadian disruption impairs bile acid and lipid homeostasis in mice. Cell Mol Gastroenterol Hepatol 1:664-677.

Fleet JC, Gliniak C, Zhang Z, Xue Y, Smith KB, McCreedy R, and Adedokun SA (2008) Serum metabolite profiles and target tissue gene expression define the effect of cholecalciferol intake on calcium metabolism in rats and mice. J Nutr 138:1114-1120.

Florentin M, Kostapanos MS, Kei A, and Elisaf MS (2014) Emerging drugs for hyperlipidaemia: an update. Expert Opin Emerg Drugs 19:471-488.

185

Folch J, Lees M, and Sloane Stanley GH (1957) A simple method for the isolation and purification of total lipides from animal tissues. J Biol Chem 226:497-509.

Foresta C, Strapazzon G, De Toni L, Perilli L, Di Mambro A, Muciaccia B, Sartori L, and Selice R (2011) Bone mineral density and testicular failure: evidence for a role of vitamin D 25- hydroxylase in human testis. J Clin Endocrinol Metab 96:E646-652.

Frazao JM, Elangovan L, Maung HM, Chesney RW, Acchiardo SR, Bower JD, Kelley BJ, Rodriguez HJ, Norris KC, Robertson JA, Levine BS, Goodman WG, Gentile D, Mazess RB, Kyllo DM, Douglass LL, Bishop CW, and Coburn JW (2000) Intermittent doxercalciferol (1α- hydroxyvitamin D2) therapy for secondary hyperparathyroidism. Am J Kidney Dis 36:550- 561.

Gascon-Barre M, Demers C, Ghrab O, Theodoropoulos C, Lapointe R, Jones G, Valiquette L, and Menard D (2001) Expression of CYP27A, a gene encoding a vitamin D-25 hydroxylase in human liver and kidney. Clin Endocrinol (Oxf) 54:107-115.

Gascon-Barre M, Demers C, Mirshahi A, Neron S, Zalzal S, and Nanci A (2003) The normal liver harbors the vitamin D nuclear receptor in nonparenchymal and biliary epithelial cells. Hepatology 37:1034-1042.

Ghanei L, Ziaee A, Rostami P, Oveisi S, Esmailzadehha N, Kazemifar AM, and Zargar A (2015) Association of serum 25-hydroxyvitamin D levels and vitamin D dietary intake with metabolic syndrome: a case control study. J Res Health Sci 15:32-36.

Gloth FM, 3rd and Tobin JD (1995) Vitamin D deficiency in older people. J Am Geriatr Soc 43:822- 828.

Goff JP, Reinhardt TA, Engstrom GW, and Horst RL (1992) Effect of dietary calcium or phosphorus restriction and 1,25-dihydroxyvitamin D administration on rat intestinal 24-hydroxylase. Endocrinology 131:101-104.

Goodwin B, Jones SA, Price RR, Watson MA, McKee DD, Moore LB, Galardi C, Wilson JG, Lewis MC, Roth ME, Maloney PR, Willson TM, and Kliewer SA (2000) A regulatory cascade of the nuclear receptors FXR, SHP-1, and LRH-1 represses bile acid biosynthesis. Mol Cell 6:517-526.

Gorham ED, Garland CF, Garland FC, Grant WB, Mohr SB, Lipkin M, Newmark HL, Giovannucci E, Wei M, and Holick MF (2005) Vitamin D and prevention of colorectal cancer. J Steroid Biochem Mol Biol 97:179-194.

Grimes DS, Hindle E, and Dyer T (1996) Sunlight, cholesterol and coronary heart disease. QJM 89:579-589.

Grostern RJ, Bryar PJ, Zimbric ML, Darjatmoko SR, Lissauer BJ, Lindstrom MJ, Lokken JM, Strugnell SA, and Albert DM (2002) Toxicity and dose-response studies of 1α- hydroxyvitamin D2 in a retinoblastoma xenograft model. Arch Ophthalmol 120:607-612.

186

Gupta AK, Sexton RC, and Rudney H (1989) Effect of vitamin D3 derivatives on cholesterol synthesis and HMG-CoA reductase activity in cultured cells. J Lipid Res 30:379-386.

Gupta GK, Agrawal T, Rai V, Del Core MG, Hunter WJ, 3rd, and Agrawal DK (2016) Vitamin D supplementation reduces intimal hyperplasia and restenosis following coronary intervention in atherosclerotic swine. PLoS One 2016;11:e0156857.

Gupta RP, Hollis BW, Patel SB, Patrick KS, and Bell NH (2004) CYP3A4 is a human microsomal vitamin D 25-hydroxylase. J Bone Miner Res 19:680-688.

Halloran BP and Castro ME (1989) Vitamin D kinetics in vivo: effect of 1,25-dihydroxyvitamin D administration. Am J Physiol 256:E686-691.

Han S and Chiang JY (2009) Mechanism of vitamin D receptor inhibition of cholesterol 7α- hydroxylase gene transcription in human hepatocytes. Drug Metab Dispos 37:469-478.

Han S, Li T, Ellis E, Strom S, and Chiang JY (2010) A novel bile acid-activated vitamin D receptor signaling in human hepatocytes. Mol Endocrinol 24:1151-1164.

Hawes JE, Tesic D, Whitehouse AJ, Zosky GR, Smith JT, and Wyrwoll CS (2015) Maternal vitamin D deficiency alters fetal brain development in the BALB/c mouse. Behav Brain Res 286:192- 200.

Hedlund TE, Moffatt KA, and Miller GJ (1996) Vitamin D receptor expression is required for growth modulation by 1α,25-dihydroxyvitamin D3 in the human prostatic carcinoma cell line ALVA- 31. J Steroid Biochem Mol Biol 58:277-288.

Heikkinen AM, Tuppurainen MT, Niskanen L, Komulainen M, Penttila I, and Saarikoski S (1997) Long-term vitamin D3 supplementation may have adverse effects on serum lipids during postmenopausal hormone replacement therapy. Eur J Endocrinol 137:495-502.

Henry HL (2001) The 25(OH)D3/1α,25(OH)2D3-24R-hydroxylase: a catabolic or biosynthetic enzyme? Steroids 66:391-398.

Hershberger PA, McGuire TF, Yu WD, Zuhowski EG, Schellens JH, Egorin MJ, Trump DL, and Johnson CS (2002) Cisplatin potentiates 1,25-dihydroxyvitamin D3-induced apoptosis in association with increased mitogen-activated protein kinase kinase kinase 1 (MEKK-1) expression. Mol Cancer Ther 1:821-829.

Hershberger PA, Yu WD, Modzelewski RA, Rueger RM, Johnson CS, and Trump DL (2001) Calcitriol (1,25-dihydroxycholecalciferol) enhances paclitaxel antitumor activity in vitro and in vivo and accelerates paclitaxel-induced apoptosis. Clin Cancer Res 7:1043-1051.

Hildmann B, Storelli C, Danisi G, and Murer H (1982) Regulation of Na+-Pi cotransport by 1,25- dihydroxyvitamin D3 in rabbit duodenal brush-border membrane. Am J Physiol 242:G533- 539.

187

Hoenderop JG, Muller D, Van Der Kemp AW, Hartog A, Suzuki M, Ishibashi K, Imai M, Sweep F, Willems PH, Van Os CH, and Bindels RJ (2001) Calcitriol controls the epithelial calcium channel in kidney. J Am Soc Neprhol 12:1342-1349.

Hoenderop JG, Nilius B, and Bindels RJ (2005) Calcium absorption across epithelia. Physiol Rev 85:373-422.

Hoenderop JG, van der Kemp AW, Hartog A, van de Graaf SF, van Os CH, Willems PH, and Bindels 2+ RJ (1999) Molecular identification of the apical Ca channel in 1,25-dihydroxyvitamin D3- responsive epithelia. J Biol Chem 274:8375-8378.

Hofmann AF (1999) The continuing importance of bile acids in liver and intestinal disease. Arch Intern Med 159:2647-2658.

Hogarth CA, Roy A, and Ebert DL (2003) Genomic evidence for the absence of a functional cholesteryl ester transport protein gene in mice and rats. Comp Biochem Physiol B Biochem Mol Biol 135:219-229.

Holick MF, Schnoes HK, DeLuca HF, Gray RW, Boyle IT, and Suda T (1972) Isolation and identification of 24,25-dihydroxycholecalciferol, a metabolite of vitamin D made in the kidney. Biochemistry 11:4251-4255.

Hollander D, Muralidhara KS, and Zimmerman A (1978) Vitamin D3 intestinal absorption in vivo: influence of fatty acids, bile salts, and perfusate pH on absorption. Gut 19:267-272.

Hollis BW (1996) Assessment of vitamin D nutritional and hormonal status: what to measure and how to do it. Calcif Tissue Int 58:4-5.

Honjo Y, Sasaki S, Kobayashi Y, Misawa H, and Nakamura H (2006) 1,25-dihydroxyvitamin D3 and its receptor inhibit the chenodeoxycholic acid-dependent transactivation by farnesoid X receptor. J Endocrinol 188:635-643.

Honma Y, Hozumi M, Abe E, Konno K, Fukushima M, Hata S, Nishii Y, DeLuca HF, and Suda T (1983) 1α,25-Dihydroxyvitamin D3 and 1α-hydroxyvitamin D3 prolong survival time of mice inoculated with myeloid leukemia cells. Proc Natl Acad Sci U S A 80:201-204.

Hsu CH, Patel S, Young EW, and Simpson RU (1987) Production and degradation of calcitriol in renal failure rats. Am J Physiol 253:F1015-1019.

Inagaki T, Choi M, Moschetta A, Peng L, Cummins CL, McDonald JG, Luo G, Jones SA, Goodwin B, Richardson JA, Gerard RD, Repa JJ, Mangelsdorf DJ, and Kliewer SA (2005) Fibroblast growth factor 15 functions as an enterohepatic signal to regulate bile acid homeostasis. Cell Metab 2:217-225.

IOM (2011) Dietary Reference Intakes for Calcium and Vitamin D. The National Academies Press, Washington, DC.

188

Ishizawa M, Matsunawa M, Adachi R, Uno S, Ikeda K, Masuno H, Shimizu M, Iwasaki K, Yamada S, and Makishima M (2008) Lithocholic acid derivatives act as selective vitamin D receptor modulators without inducing hypercalcemia. J Lipid Res 49:763-772.

Itoh S, Yoshimura T, Iemura O, Yamada E, Tsujikawa K, Kohama Y, and Mimura T (1995) Molecular cloning of 25-hydroxyvitamin D3 24-hydroxylase (Cyp-24) from mouse kidney: its inducibility by vitamin D3. Biochim Biophys Acta 1264:26-28.

Jiang W, Miyamoto T, Kakizawa T, Nishio SI, Oiwa A, Takeda T, Suzuki S, and Hashizume K (2006) Inhibition of LXRα signaling by vitamin D receptor: possible role of VDR in bile acid synthesis. Biochem Biophys Res Commun 351:176-184.

Jolliffe DA, Walton RT, Griffiths CJ, and Martineau AR (2015) Single nucleotide polymorphisms in the vitamin D pathway associating with circulating concentrations of vitamin D metabolites and non-skeletal health outcomes: Review of genetic association studies. J Steroid Biochem Mol Biol [Epub doi: 10.1016/j.jsbmb.2015.12.007].

Jones G (2008) Pharmacokinetics of vitamin D toxicity. Am J Clin Nutr 88:582S-586S.

Jones G, Strugnell SA, and DeLuca HF (1998) Current understanding of the molecular actions of vitamin D. Physiol Rev 78:1193-1231.

Jones G, Vriezen D, Lohnes D, Palda V, and Edwards NS (1987) Side-chain hydroxylation of vitamin D3 and its physiological implications. Steroids 49:29-53.

Joshi R. Hypercalcemia due to hypervitaminosis D: report of seven patients (2009) J Trop Pediatrics 55:396-398.

Jouni ZE, Winzerling JJ, and McNamara DJ (1995) 1,25-Dihydroxyvitamin D3-induced HL-60 macrophages: regulation of cholesterol and LDL metabolism. Atherosclerosis 117:125-138.

Kakara M, Nomura H, Fukae M, Gotanda K, Hirota T, Matsubayashi S, Shimomura H, Hirakawa M, and Ieiri I (2014) Population pharmacodynamic analysis of LDL-cholesterol lowering effects by statins and co-medications based on electronic medical records. Br J Clin Pharmacol 78:824-835.

Kamei Y, Kawada T, Fukuwatari T, Ono T, Kato S, and Sugimoto E (1995) Cloning and sequencing of the gene encoding the mouse vitamin D receptor. Gene 152:281-282.

Kannel WB, Castelli WP, and Gordon T (1979) Cholesterol in the prediction of atherosclerotic disease. New perspectives based on the Framingham study. Ann Intern Med 90:85-91.

Karhapaa P, Pihlajamaki J, Porsti I, Kastarinen M, Mustonen J, Niemela O, and Kuusisto J (2010) Diverse associations of 25-hydroxyvitamin D and 1,25-dihydroxy-vitamin D with dyslipidaemias. J Intern Med 268:604-610.

189

Karnauskas AJ, van Leeuwen JP, van den Bemd GJ, Kathpalia PP, DeLuca HF, Bushinsky DA, and Favus MJ (2005) Mechanism and function of high vitamin D receptor levels in genetic hypercalciuric stone-forming rats. J Bone Miner Res 20:447-454.

Khan AA, Chow EC, van Loenen-Weemaes AM, Porte RJ, Pang KS, and Groothuis GM (2009) Comparison of effects of VDR versus PXR, FXR and GR ligands on the regulation of CYP3A isozymes in rat and human intestine and liver. Eur J Pharm Sci 37:115-125.

Kilkenny C, Browne W, Cuthill IC, Emerson M, Altman DG, and Group NCRRGW (2010) Animal research: reporting in vivo experiments: the ARRIVE guidelines. Br J Pharmacol 160:1577- 1579.

Kim YC, Kim IB, Noh CK, Quach HP, Yoon IS, Chow EC, Kim M, Jin HE, Cho KH, Chung SJ, Pang KS, and Maeng HJ (2014) Effects of 1α,25-dihydroxyvitamin D3, the natural vitamin D receptor ligand, on the pharmacokinetics of cefdinir and cefadroxil, organic anion transporter substrates, in rat. J Pharm Sci 103:3793-3805.

Kimura Y, Nakayama M, Kuriyama S, Watanabe S, Kawaguchi Y, and Sakai O (1991) Pharmacokinetics of active vitamins D3, 1α-hydroxyvitamin D3 and 1α,25-dihydroxyvitamin D3 in patients on chronic hemodialysis. Clin Nephrol 35:72-77.

Kissmeyer AM and Binderup L (1991) Calcipotriol (MC 903): pharmacokinetics in rats and biological activities of metabolites. A comparative study with 1,25(OH)2D3 Biochem Pharmacol 41:1601-1606.

Kong M, Zhu L, Bai L, Zhang X, Chen Y, Liu S, Zheng S, Pandol SJ, Han YP, and Duan Z (2014) Vitamin D deficiency promotes nonalcoholic steatohepatitis through impaired enterohepatic circulation in animal model. Am J Physiol Gastrointest Liver Physiol 307:G883-893.

Kragballe K (1995) Calcipotriol: a new drug for topical psoriasis treatment. Pharmacol Toxicol 77:241-246.

Kumar R, Schnoes HK, and DeLuca HF (1978) Rat intestinal 25-hydroxyvitamin D3- and 1α,25- dihydroxyvitamin D3-24-hydroxylase. J Biol Chem 253:3804-3809.

Kutuzova GD and DeLuca HF (2007) 1,25-Dihydroxyvitamin D3 regulates genes responsible for detoxification in intestine. Toxicol Appl Pharmacol 218:37-44.

Lawson DE, Sedrani SH, and Douglas J (1986) Interrelationships in rats of tissue pools of cholecalciferol and 25-hydroxycholecalciferol formed in u.v. light. Biochem J 233:535-540.

Lee HJ, Muindi JR, Tan W, Hu Q, Wang D, Liu S, Wilding GE, Ford LA, Sait SN, Block AW, Adjei AA, Barcos M, Griffiths EA, Thompson JE, Wang ES, Johnson CS, Trump DL, and Wetzler M (2014) Low 25(OH) vitamin D3 levels are associated with adverse outcome in newly diagnosed, intensively treated adult acute myeloid leukemia. Cancer 120:521-529.

Lee JH, O'Keefe JH, Bell D, Hensrud DD, and Holick MF (2008) Vitamin D deficiency an important, common, and easily treatable cardiovascular risk factor? J Am Coll Cardiol 52:1949-1956.

190

Lee M, Ebert JR, Kadakia MP, Zhang J, and Czerwinski SA (2016) Inverse associations between cardiometabolic risk factors and 25-hydroxyvitamin D in obese American children and adolescents. Am J Hum Biol [Epub doi: 10.1002/ajhb.22863].

Lee YK, Dell H, Dowhan DH, Hadzopoulou-Cladaras M, and Moore DD (2000) The orphan nuclear receptor SHP inhibits hepatocyte nuclear factor 4 and retinoid X receptor transactivation: two mechanisms for repression. Mol Cell Biol 20:187-195.

Lefebvre P, Cariou B, Lien F, Kuipers F, and Staels B (2009) Role of bile acids and bile acid receptors in metabolic regulation. Physiol Rev 89:147-191.

Lemay J, Demers C, Hendy GN, Delvin EE, and Gascon-Barre M (1995) Expression of the 1,25- dihydroxyvitamin D3-24-hydroxylase gene in rat intestine: response to calcium, vitamin D3 and calcitriol administration in vivo. J Bone Miner Res 10:1148-1157.

Lemon BD and Freedman LP (1996) Selective effects of ligands on vitamin D3 receptor- and retinoid X receptor-mediated gene activation in vivo. Mol Cell Biol 16:1006-1016.

Levy G (1964) Relationship between elimination rate of drugs and rate of decline of their pharmacologic effects. J Pharm Sci 53:342-343.

Li-Hawkins J, Gafvels M, Olin M, Lund EG, Andersson U, Schuster G, Bjorkhem I, Russell DW, and Eggertsen G (2002) Cholic acid mediates negative feedback regulation of bile acid synthesis in mice. J Clin Invest 110:1191-1200.

Li S, He Y, Lin S, Hao L, Ye Y, Lv L, Sun Z, Fan H, Shi Z, Li J, Feng R, Na L, Wang Y, Li Y, and Sun C (2016) Increase of circulating cholesterol in vitamin D deficiency is linked to reduced vitamin D receptor activity via the Insig-2/SREBP-2 pathway. Mol Nutr Food Res 60:798- 809.

Longenecker CT, Hileman CO, Carman TL, Ross AC, Seydafkan S, Brown TT, Labbato DE, Storer N, Tangpricha V, and McComsey GA (2012) Vitamin D supplementation and endothelial function in vitamin D deficient HIV-infected patients: a randomized placebo-controlled trial. Antivir Ther 17:613-621.

Lowry OH, Rosebrough NJ, Farr AL, and Randall RJ (1951) Protein measurement with the Folin phenol reagent. J Biol Chem 193:265-275.

Lu TT, Makishima M, Repa JJ, Schoonjans K, Kerr TA, Auwerx J, and Mangelsdorf DJ (2000) Molecular basis for feedback regulation of bile acid synthesis by nuclear receptors. Mol Cell 6:507-515.

Maeng HJ, Chapy H, Zaman S, and Pang KS (2012) Effects of 1α,25-dihydroxyvitamin D3 on transport and metabolism of adefovir dipivoxil and its metabolites in Caco-2 cells. Eur J Pharm Sci 46:149-166.

Mager DE and Jusko WJ (2001) Pharmacodynamic modeling of time-dependent transduction systems. Clin Pharmacol Ther 70:210-216.

191

Mager DE, Woo S, and Jusko WJ (2009) Scaling pharmacodynamics from in vitro and preclinical animal studies to humans. Drug Metab Pharmacokinet 24:16-24.

Mager DE, Wyska E, and Jusko WJ (2003) Diversity of mechanism-based pharmacodynamic models. Drug Metab Dispos 31:510-518.

Mahmoudi T, Mohebbi SR, Pourhoseingholi MA, Fatemi SR, and Zali MR (2010) Vitamin D receptor gene ApaI polymorphism is associated with susceptibility to colorectal cancer. Dig Dis Sci 55:2008-2013.

Makin G, Lohnes D, Byford V, Ray R, and Jones G (1989) Target cell metabolism of 1,25- dihydroxyvitamin D3 to calcitroic acid. Evidence for a pathway in kidney and bone involving 24-oxidation. Biochem J 262:173-180.

Makishima M, Lu TT, Xie W, Whitfield GK, Domoto H, Evans RM, Haussler MR, and Mangelsdorf DJ (2002) Vitamin D receptor as an intestinal bile acid sensor. Science 296:1313-1316.

Makishima M, Okamoto AY, Repa JJ, Tu H, Learned RM, Luk A, Hull MV, Lustig KD, Mangelsdorf DJ, and Shan B (1999) Identification of a nuclear receptor for bile acids. Science 284:1362- 1365.

Mangelsdorf DJ, Thummel C, Beato M, Herrlich P, Schutz G, Umesono K, Blumberg B, Kastner P, Mark M, Chambon P, and Evans RM (1995) The nuclear receptor superfamily: the second decade. Cell 83:835-839.

Martin MJ, Hulley SB, Browner WS, Kuller LH, and Wentworth D (1986) Serum cholesterol, blood pressure, and mortality: implications from a cohort of 361,662 men. Lancet 2:933-936.

Masuda S, Byford V, Arabian A, Sakai Y, Demay MB, St-Arnaud R, and Jones G (2005) Altered pharmacokinetics of 1α,25-dihydroxyvitamin D3 and 25-hydroxyvitamin D3 in the blood and tissues of the 25-hydroxyvitamin D-24-hydroxylase (Cyp24a1) null mouse. Endocrinology 146:825-834.

McCarthy TC, Li X, and Sinal CJ (2005) Vitamin D receptor-dependent regulation of colon multidrug resistance-associated protein 3 gene expression by bile acids. J Biol Chem 280:23232-23242.

Meyer MB, Watanuki M, Kim S, Shevde NK, and Pike JW (2006) The human transient receptor potential vanilloid type 6 distal promoter contains multiple vitamin D receptor binding sites that mediate activation by 1,25-dihydroxyvitamin D3 in intestinal cells. Mol Endocrinol 20:1447-1461.

Meyer MB, Zella LA, Nerenz RD, and Pike JW (2007) Characterizing early events associated with the activation of target genes by 1,25-dihydroxyvitamin D3 in mouse kidney and intestine in vivo. J Biol Chem 282:22344-22352.

192

Miao J, Xiao Z, Kanamaluru D, Min G, Yau PM, Veenstra TD, Ellis E, Strom S, Suino-Powell K, Xu HE, and Kemper JK (2009) Bile acid signaling pathways increase stability of small heterodimer partner (SHP) by inhibiting ubiquitin-proteasomal degradation. Genes Dev 23:986-996.

Mithal A, Wahl DA, Bonjour JP, Burckhardt P, Dawson-Hughes B, Eisman JA, El-Hajj Fuleihan G, Josse RG, Lips P, Morales-Torres J (2009) Global vitamin D status and determinants of hypovitaminosis D. Osteoporosis Int 20:1807-1820.

Muindi JR, Modzelewski RA, Peng Y, Trump DL, and Johnson CS (2004) Pharmacokinetics of 1α,25-dihydroxyvitamin D3 in normal mice after systemic exposure to effective and safe antitumor doses. Oncology 66:62-66.

Muindi JR, Peng Y, Potter DM, Hershberger PA, Tauch JS, Capozzoli MJ, Egorin MJ, Johnson CS, and Trump DL (2002) Pharmacokinetics of high-dose oral calcitriol: results from a phase 1 trial of calcitriol and paclitaxel. Clin Pharmacol Ther 72:648-659.

Muindi JR, Yu WD, Ma Y, Engler KL, Kong RX, Trump DL, and Johnson CS (2010) CYP24A1 inhibition enhances the antitumor activity of calcitriol. Endocrinology 151:4301-4312.

Mun MJ, Kim TH, Hwang JY, and Jang WC (2015) Vitamin D receptor gene polymorphisms and the risk for female reproductive cancers: A meta-analysis. Maturitas 81:256-265.

Muscogiuri G, Sorice GP, Prioletta A, Policola C, Della Casa S, Pontecorvi A, and Giaccari A (2010) 25-Hydroxyvitamin D concentration correlates with insulin-sensitivity and BMI in obesity. Obesity 18:1906-1910.

Navarro-Valverde C, Sosa-Henriquez M, Alhambra-Exposito MR, and Quesada-Gomez JM (2016) Vitamin D and calcidiol are not equipotent. J Steroid Biochem Mol Biol [Epub doi: 10.1016/j.jsbmb.2016.01.014].

Nishida S, Ozeki J, and Makishima M (2009) Modulation of bile acid metabolism by 1α- hydroxyvitamin D3 administration in mice. Drug Metab Dispos 37:2037-2044.

Noshiro M, Nishimoto M, and Okuda K (1990) Rat liver cholesterol 7α-hydroxylase. Pretranslational regulation for circadian rhythm. J Biol Chem 265:10036-10041.

Nutchey BK, Kaplan JS, Dwivedi PP, Omdahl JL, Ferrante A, May BK, and Hii CS (2005) Molecular action of 1,25-dihydroxyvitamin D3 and phorbol ester on the activation of the rat cytochrome P450C24 (CYP24) promoter: role of MAP kinase activities and identification of an important transcription factor binding site. Biochem J 389:753-762.

Nykjaer A, Dragun D, Walther D, Vorum H, Jacobsen C, Herz J, Melsen F, Christensen EI, and Willnow TE (1999) An endocytic pathway essential for renal uptake and activation of the steroid 25-(OH) vitamin D3. Cell 96:507-515.

193

Ogura M, Nishida S, Ishizawa M, Sakurai K, Shimizu M, Matsuo S, Amano S, Uno S, and Makishima M (2009) Vitamin D3 modulates the expression of bile acid regulatory genes and represses inflammation in bile duct-ligated mice. J Pharmacol Exp Ther 328:564-570.

Oliveira RM, Novaes JF, Azeredo LM, Candido AP, and Leite IC (2014) Association of vitamin D insufficiency with adiposity and metabolic disorders in Brazilian adolescents. Public Health Nutr 17:787-794.

Olson EB, Jr., Knutson JC, Bhattacharyya MH, and DeLuca HF (1976) The effect of hepatectomy on the synthesis of 25-hydroxyvitamin D3. J Clin Invest 57:1213-1220.

Ooi LL, Zhou H, Kalak R, Zheng Y, Conigrave AD, Seibel MJ, and Dunstan CR (2010) Vitamin D deficiency promotes human breast cancer growth in a murine model of bone metastasis. Cancer Res 70:1835-1844.

Pang KS (2003) Modeling of intestinal drug absorption: roles of transporters and metabolic enzymes (for the Gillette Review Series). Drug Metab Dispos 31:1507-1519.

Pang KS and Chow EC (2012) Commentary: theoretical predictions of flow effects on intestinal and systemic availability in physiologically based pharmacokinetic intestine models: the traditional model, segregated flow model, and QGut model. Drug Metab Dispos 40:1869- 1877.

Panierakis C, Goulielmos G, Mamoulakis D, Petraki E, Papavasiliou E, and Galanakis E (2009) Vitamin D receptor gene polymorphisms and susceptibility to type 1 diabetes in Crete, Greece. Clin Immunol 133:276-281.

Park S, Kim da S, and Kang S (2016) Vitamin D deficiency impairs glucose-stimulated insulin secretion and increases insulin resistance by reducing PPARγ expression in nonobese Type 2 diabetic rats. J Nutr Biochem 27:257-265.

Patel R, Patel M, Tsai R, Lin V, Bookout AL, Zhang Y, Magomedova L, Li T, Chan JF, Budd C, Mangelsdorf DJ, and Cummins CL (2011) LXRβ is required for glucocorticoid-induced hyperglycemia and hepatosteatosis in mice. J Clin Invest 121:431-441.

Patwardhan VG, Khadilkar AV, Chiplonkar SA, Mughal ZM, and Khadilkar VV (2015) Varying relationship between 25-hydroxy-vitamin D, high density lipoprotein cholesterol, and serum 7-dehydrocholesterol reductase with sunlight exposure. J Clin Lipidol 9:652-657.

Pekkanen MP, Ukkola O, Hedberg P, Piira OP, Lepojarvi S, Lumme J, Tulppo MP, and Huikuri HV (2015) Serum 25-hydroxyvitamin D is associated with major cardiovascular risk factors and cardiac structure and function in patients with coronary artery disease. Nutr Metab Cardiovasc Dis 25:471-478.

Peleg S (2005) Molecular Basis for Differential Action of Vitamin D Analogs, in: Vitamin D (Feldman D ed), pp 1471-1488, Elsevier, New York.

194

Perna L, Butterbach K, Haug U, Schottker B, Muller H, Arndt V, Holleczek B, Burwinkel B, and Brenner H (2013) Vitamin D receptor genotype rs731236 (Taq1) and breast cancer prognosis. Cancer Epidemiol Biomarkers Prev 22:437-442.

Petersen RA, Dalskov SM, Sorensen LB, Hjorth MF, Andersen R, Tetens I, Krarup H, Ritz C, Astrup A, Michaelsen KF, Molgaard C, and Damsgaard CT (2015) Vitamin D status is associated with cardiometabolic markers in 8-11-year-old children, independently of body fat and physical activity. Br J Nutr 114:1647-1655.

Pfrunder A, Gutmann H, Beglinger C, and Drewe J (2003) Gene expression of CYP3A4, ABC- transporters (MDR1 and MRP1-MRP5) and hPXR in three different human colon carcinoma cell lines. J Pharm Pharmcaol 55:59-66.

Ponda MP, Dowd K, Finkielstein D, Holt PR, and Breslow JL (2012) The short-term effects of vitamin D repletion on cholesterol: a randomized, placebo-controlled trial. Arterioscler Thromb Vasc Biol 32:2510-2515.

Prabhu AV, Luu W, Sharpe LJ, and Brown AJ (2016) Cholesterol-mediated degradation of 7- dehydrocholesterol reductase switches the balance from cholesterol to vitamin D synthesis. J Biol Chem 291:8363-8373.

Prentice A (2013) Nutritional rickets around the world. J Steroid Biochem Mol Biol 136:201-206.

Prudencio J, Akutsu N, Benlimame N, Wang T, Bastien Y, Lin R, Black MJ, Alaoui-Jamali MA, and White JH (2001) Action of low calcemic 1α,25-dihydroxyvitamin D3 analogue EB1089 in head and neck squamous cell carcinoma. J Natl Cancer Inst 93:745-753.

Quach HP, Yang QJ, Chow EC, Mager DE, Hoi SY, and Pang KS (2015) PKPD modelling to predict altered disposition of 1α,25-dihydroxyvitamin D3 in mice due to dose-dependent regulation of CYP27B1 on synthesis and CYP24A1 on degradation. Br J Pharmacol 172:3611-3626.

Quintao E, Grundy SM, and Ahrens EH, Jr. (1971) Effects of dietary cholesterol on the regulation of total body cholesterol in man. J Lipid Res 12:233-247.

Rahimi-Ardabili H, Pourghassem Gargari B, and Farzadi L (2013) Effects of vitamin D on cardiovascular disease risk factors in polycystic ovary syndrome women with vitamin D deficiency. J Endocrinol Invest 36:28-32.

Ramakrishnan V, Yang QJ, Quach HP, Cao Y, Chow EC, Mager DE, and Pang KS (2016) Physiologically-based pharmacokinetic-pharmacodynamic modeling of 1α,25- dihydroxyvitamin D3 in mice. Drug Metab Dispos 44:189-208.

Ramiro-Lozano JM and Calvo-Romero JM (2015) Effects on lipid profile of supplementation with vitamin D in type 2 diabetic patients with vitamin D deficiency. Ther Adv Endocrinol Metab 6:245-248.

195

Ramnath N, Daignault-Newton S, Dy GK, Muindi JR, Adjei A, Elingrod VL, Kalemkerian GP, Cease KB, Stella PJ, Brenner DE, Troeschel S, Johnson CS, and Trump DL (2013) A phase I/II pharmacokinetic and pharmacogenomic study of calcitriol in combination with cisplatin and docetaxel in advanced non-small-cell lung cancer. Cancer Chemother Pharmacol 71:1173- 1182.

Rassnick KM, Muindi JR, Johnson CS, Balkman CE, Ramnath N, Yu WD, Engler KL, Page RL, and Trump DL (2008) In vitro and in vivo evaluation of combined calcitriol and cisplatin in dogs with spontaneously occurring tumors. Cancer Chemother Pharmacol 62:881-891.

Reid IR and Bolland MJ (2014) Skeletal and nonskeletal effects of vitamin D: is vitamin D a tonic for bone and other tissues? Osteoporosis Int 25:2347-2357.

Reschly EJ and Krasowski MD (2006) Evolution and function of the NR1I nuclear hormone receptor subfamily (VDR, PXR, and CAR) with respect to metabolism of xenobiotics and endogenous compounds. Curr Drug Metab 7:349-365.

Revelli A, Massobrio M, and Tesarik J (1998) Nongenomic effects of 1α,25-dihydroxyvitamin D3. Trends Endocrinol Metab 9:419-427.

Rochel N, Wurtz JM, Mitschler A, Klaholz B, and Moras D (2000) The crystal structure of the nuclear receptor for vitamin D bound to its natural ligand. Mol Cell 5:173-179.

Roomi MA, Lone KP, and Madassar A (2014) Vitamin D and cardiometabolic risk factors in adult non-diabetic offspring of type 2 diabetic parents. J Pak Med Assoc 64:1229-1234.

Roth CL, Elfers CT, Figlewicz DP, Melhorn SJ, Morton GJ, Hoofnagle A, Yeh MM, Nelson JE, and Kowdley KV (2012) Vitamin D deficiency in obese rats exacerbates nonalcoholic fatty liver disease and increases hepatic resistin and Toll-like receptor activation. Hepatology 55:1103- 1111.

Rusconi RE, De Cosmi V, Gianluca G, Giavoli C, and Agostoni C (2015) Vitamin D insufficiency in obese children and relation with lipid profile. Int J Food Sci Nutr 66:132-134.

Saeki M, Kurose K, Tohkin M, and Hasegawa R (2008) Identification of the functional vitamin D response elements in the human MDR1 gene. Biochem Pharmacol 76:531-542.

Safadi FF, Thornton P, Magiera H, Hollis BW, Gentile M, Haddad JG, Liebhaber SA, and Cooke NE (1999) Osteopathy and resistance to vitamin D toxicity in mice null for vitamin D binding protein. J Clin Invest 103:239-251.

Salehpour A, Shidfar F, Hosseinpanah F, Vafa M, Razaghi M, Hoshiarrad A, and Gohari M (2012) Vitamin D3 and the risk of CVD in overweight and obese women: a randomised controlled trial. Br J Nutr 108:1866-1873.

Salusky IB, Goodman WG, Horst R, Segre GV, Kim L, Norris KC, Adams JS, Holloway M, Fine RN, and Coburn JW (1990) Pharmacokinetics of calcitriol in continuous ambulatory and cycling peritoneal dialysis patients. Am J Kidney Dis 16:126-132.

196

Salusky IB, Goodman WG, Horst R, Segre GV, Norris KC, and Coburn JW (1988) Plasma kinetics of intravenous calcitriol in normal and dialysed subjects and acute effect on serum nPTH levels, in: Vitamin D: Molecular, Cellular and Clinical Endocrinology Proceedings of the 7th Workshop on Vitamin D (Norman AW ed), pp 781-782, Walter de Gruyter, New York.

Sandgren ME, Bronnegard M, and DeLuca HF (1991) Tissue distribution of the 1,25- dihydroxyvitamin D3 receptor in the male rat. Biochem Biophys Res Commun 181:611-616.

Sarkinen BE (2011) Vitamin D deficiency & cardiovascular disease. Nurse Pract 36:46-53.

Schmidt DR, Holmstrom SR, Fon Tacer K, Bookout AL, Kliewer SA, and Mangelsdorf DJ (2010) Regulation of bile acid synthesis by fat-soluble vitamins A and D. J Biol Chem 285:14486- 14494.

Schmidt N, Brandsch C, Schutkowski A, Hirche F, and Stangl GI (2014) Dietary vitamin D inadequacy accelerates calcification and osteoblast-like cell formation in the vascular system of LDL receptor knockout and wild-type mice. J Nutr 144:638-646.

Schmiedlin-Ren P, Thummel KE, Fisher JM, Paine MF, Lown KS, and Watkins PB (1997) Expression of enzymatically active CYP3A4 by Caco-2 cells grown on extracellular matrix- coated permeable supports in the presence of 1α,25-dihydroxyvitamin D3. Mol Pharmacol 51:741-754.

Schmiedlin-Ren P, Thummel KE, Fisher JM, Paine MF, and Watkins PB (2001) Induction of CYP3A4 by 1α,25-dihydroxyvitamin D3 is human cell line-specific and is unlikely to involve pregnane X receptor. Drug Metab Dispos 29:1446-1453.

Schnatz PF, Jiang X, Vila-Wright S, Aragaki AK, Nudy M, O'Sullivan DM, Jackson R, LeBlanc E, Robinson JG, Shikany JM, Womack CR, Martin LW, Neuhouser ML, Vitolins MZ, Song Y, Kritchevsky S, and Manson JE (2014) Calcium/vitamin D supplementation, serum 25- hydroxyvitamin D concentrations, and cholesterol profiles in the Women's Health Initiative calcium/vitamin D randomized trial. Menopause 21:823-833.

Schwartz JB (2009) Effects of vitamin D supplementation in atorvastatin-treated patients: a new drug interaction with an unexpected consequence. Clin Pharmacol Ther 85:198-203.

Sertznig P, Dunlop T, Seifert M, Tilgen W, and Reichrath J (2009) Cross-talk between vitamin D receptor (VDR)- and peroxisome proliferator-activated receptor (PPAR)-signaling in melanoma cells. Anticancer Res 29:3647-3658.

Sharma A and Jusko WJ (1996) Characterization of four basic models of indirect pharmacodynamic responses. J Pharmacokinet Biopharm 24:611-635.

Shinki T, Jin CH, Nishimura A, Nagai Y, Ohyama Y, Noshiro M, Okuda K, and Suda T (1992) Parathyroid hormone inhibits 25-hydroxyvitamin D3-24-hydroxylase mRNA expression stimulated by 1α,25-dihydroxyvitamin D3 in rat kidney but not in intestine. J Biol Chem 267:13757-13762.

197

Slatopolsky E, Weerts C, Thielan J, Horst R, Harter H, and Martin KJ (1984) Marked suppression of secondary hyperparathyroidism by intravenous administration of 1,25-dihydroxycholecalciferol in uremic patients. J Clin Invest 74:2136-2143.

Smith DC, Johnson CS, Freeman CC, Muindi J, Wilson JW, and Trump DL (1999) A Phase I trial of calcitriol (1,25-dihydroxycholecalciferol) in patients with advanced malignancy. Clin Cancer Res 5:1339-1345.

Song Y and Fleet JC (2007) Intestinal resistance to 1,25 dihydroxyvitamin D in mice heterozygous for the vitamin D receptor knockout allele. Endocrinology 148:1396-1402.

St Peter WL (2000) Hectorol: a new vitamin D prohormone. Nephrol Nurs J 27:67-68.

Staudinger JL, Goodwin B, Jones SA, Hawkins-Brown D, MacKenzie KI, LaTour A, Liu Y, Klaassen CD, Brown KK, Reinhard J, Willson TM, Koller BH, and Kliewer SA (2001) The nuclear receptor PXR is a lithocholic acid sensor that protects against liver toxicity. Proc Natl Acad Sci U S A 98:3369-3374.

Stavenuiter AW, Arcidiacono MV, Ferrantelli E, Keuning ED, Vila Cuenca M, ter Wee PM, Beelen RH, Vervloet MG, and Dusso AS (2015) A novel rat model of vitamin D deficiency: safe and rapid induction of vitamin D and calcitriol deficiency without hyperparathyroidism. Biomed Res Int 2015:604275.

Strugnell S, Byford V, Makin HL, Moriarty RM, Gilardi R, LeVan LW, Knutson JC, Bishop CW, and Jones G (1995) 1α,24(S)-dihydroxyvitamin D2: a biologically active product of 1α- hydroxyvitamin D2 made in the human hepatoma, Hep3B. Biochem J 310 ( Pt 1):233-241.

Sun X, Cao ZB, Tanisawa K, Ito T, Oshima S, Ishimi Y, Tabata I, and Higuchi M (2015) Associations between the serum 25(OH)D concentration and lipid profiles in Japanese men. J Atheroscler Thromb 22:355-362.

Suzuki Y, Landowski CP, and Hediger MA (2008) Mechanisms and regulation of epithelial Ca2+ absorption in health and disease. Annu Rev Physiol 70:257-271.

Swami S, Krishnan AV, Wang JY, Jensen K, Peng L, Albertelli MA, and Feldman D (2011) Inhibitory effects of calcitriol on the growth of MCF-7 breast cancer xenografts in nude mice: selective modulation of aromatase expression in vivo. Horm Cancer 2:190-202.

Tabesh M, Azadbakht L, Faghihimani E, Tabesh M, and Esmaillzadeh A (2014) Effects of calcium- vitamin D co-supplementation on metabolic profiles in vitamin D insufficient people with type 2 diabetes: a randomised controlled clinical trial. Diabetologia 57:2038-2047.

Takemoto F, Shinki T, Yokoyama K, Inokami T, Hara S, Yamada A, Kurokawa K, and Uchida S (2003) Gene expression of vitamin D hydroxylase and megalin in the remnant kidney of nephrectomized rats. Kidney Int 64:414-420.

198

Taketani Y, Segawa H, Chikamori M, Morita K, Tanaka K, Kido S, Yamamoto H, Iemori Y, Tatsumi S, Tsugawa N, Okano T, Kobayashi T, Miyamoto K, and Takeda E (1998) Regulation of type + II renal Na -dependent inorganic phosphate transporters by 1,25-dihydroxyvitamin D3. Identification of a vitamin D-responsive element in the human NAPi-3 gene. J Biol Chem 273:14575-14581.

Takeyama K and Kato S (2011) The vitamin D3 1α-hydroxylase gene and its regulation by active vitamin D3. Biosci Biotechnol Biochem 75:208-213.

Tanaka Y and DeLuca HF (1984) Rat renal 25-hydroxyvitamin D3 1- and 24-hydroxylases: their in vivo regulation. Am J Physiol 246:E168-173.

Tangpricha V, Spina C, Yao M, Chen TC, Wolfe MM, and Holick MF (2005) Vitamin D deficiency enhances the growth of MC-26 colon cancer xenografts in Balb/c mice. J Nutr 135:2350- 2354.

Theodoropoulos C, Demers C, Delvin E, Menard D, and Gascon-Barre M (2003) Calcitriol regulates the expression of the genes encoding the three key vitamin D3 hydroxylases and the drug- metabolizing enzyme CYP3A4 in the human fetal intestine. Clin Endocrinol (Oxf) 58:489-99.

Thummel KE, Brimer C, Yasuda K, Thottassery J, Senn T, Lin Y, Ishizuka H, Kharasch E, Schuetz J, and Schuetz E (2001) Transcriptional control of intestinal cytochrome P-4503A by 1α,25- dihydroxy vitamin D3. Mol Pharmacol 60:1399-1406.

Tiosano D, Wildbaum G, Gepstein V, Verbitsky O, Weisman Y, Karin N, and Eztioni A (2013) The role of vitamin D receptor in innate and adaptive immunity: a study in hereditary vitamin D- resistant rickets patients. J Clin Endocrinol Metab 98:1685-1693.

Turunen MM, Dunlop TW, Carlberg C, and Vaisanen S (2007) Selective use of multiple vitamin D response elements underlies the 1α,25-dihydroxyvitamin D3-mediated negative regulation of the human CYP27B1 gene. Nucleic Acids Res 35:2734-2747.

Uitterlinden AG, Fang Y, Van Meurs JB, Pols HA, and Van Leeuwen JP (2004a) Genetics and biology of vitamin D receptor polymorphisms. Gene 338:143-156.

Uitterlinden AG, Fang Y, van Meurs JB, van Leeuwen H, and Pols HA (2004b) Vitamin D receptor gene polymorphisms in relation to vitamin D related disease states. J Steroid Biochem Mol Biol 89-90:187-193.

Valdivielso JM, Cannata-Andia J, Coll B, and Fernandez E (2009) A new role for vitamin D receptor activation in chronic kidney disease. Am J Physiol Renal Physiol 297:F1502-1509. van der Wulp MY, Verkade HJ, and Groen AK (2013) Regulation of cholesterol homeostasis. Mol Cell Endocrinol 368:1-16.

Verdoia M, Schaffer A, Sartori C, Barbieri L, Cassetti E, Marino P, Galasso G, and De Luca G (2014) Vitamin D deficiency is independently associated with the extent of coronary artery disease. Eur J Clin Invest 44:634-642.

199

Vieth R, Fraser D, and Kooh SW (1987) Low dietary calcium reduces 25-hydroxycholecalciferol in plasma of rats. J Nutr 117:914-918.

Vieth R, Kooh SW, Balfe JW, Rawlins M, and Tinmouth WW (1990a) Tracer kinetics and actions of oral and intraperitoneal 1,25-dihydroxyvitamin D3 administration in rats. Kidney Int 38:857- 861.

Vieth R, McCarten K, and Norwich KH (1990b) Role of 25-hydroxyvitamin D3 dose in determining rat 1,25-dihydroxyvitamin D3 production. Am J Physiol 258:E780-789.

Wagner D, Dias AG, Schnabl K, Van der Kwast T, and Vieth R (2012) Determination of 1,25- dihydroxyvitamin D concentrations in human colon tissues and matched serum samples. Anticancer Res 32:259-263.

Wagner M, Zollner G, and Trauner M (2010) Nuclear receptor regulation of the adaptive response of bile acid transporters in cholestasis. Semin Liver Dis 30:160-177.

Wang JH, Keisala T, Solakivi T, Minasyan A, Kalueff AV, and Tuohimaa P (2009) Serum cholesterol and expression of ApoAI, LXRβ and SREBP2 in vitamin D receptor knock-out mice. J Steroid Biochem Mol Biol 113:222-226.

Wang L, Hara K, Van Baaren JM, Price JC, Beecham GW, Gallins PJ, Whitehead PL, Wang G, Lu C, Slifer MA, Zuchner S, Martin ER, Mash D, Haines JL, Pericak-Vance MA, and Gilbert JR (2012) Vitamin D receptor and Alzheimer's disease: a genetic and functional study. Neurobiol Aging 33:1844 e1841-1849.

Wang TJ, Pencina MJ, Booth SL, Jacques PF, Ingelsson E, Lanier K, Benjamin EJ, D'Agostino RB, Wolf M, and Vasan RS (2008) Vitamin D deficiency and risk of cardiovascular disease. Circulation 117:503-511.

Wang XX, Jiang T, Shen Y, Santamaria H, Solis N, Arbeeny C, and Levi M (2011) Vitamin D receptor agonist doxercalciferol modulates dietary fat-induced renal disease and renal lipid metabolism. Am J Physiol Renal Physiol 300:F801-810.

Weber K, Erben RG, Rump A, and Adamski J (2001) Gene structure and regulation of the murine epithelial calcium channels ECaC1 and 2. Biochem Biophys Res Commun 289:1287-1294.

Weng S, Sprague JE, Oh J, Riek AE, Chin K, Garcia M, and Bernal-Mizrachi C (2013) Vitamin D deficiency induces high blood pressure and accelerates atherosclerosis in mice. PLoS One 8:e54625.

Wu FS, Zheng SS, Wu LJ, Teng LS, Ma ZM, Zhao WH, and Wu W (2007) Calcitriol inhibits the growth of MHCC97 heptocellular cell lines by down-modulating c-met and ERK expressions. Liver Int 27:700-707.

Xu G, Salen G, Shefer S, Ness GC, Chen TS, Zhao Z, and Tint GS (1995) Reproducing abnormal cholesterol biosynthesis as seen in the Smith-Lemli-Opitz syndrome by inhibiting the conversion of 7-dehydrocholesterol to cholesterol in rats. J Clin Invest 95:76-81.

200

Xue Y and Fleet JC (2009) Intestinal vitamin D receptor is required for normal calcium and bone metabolism in mice. Gastroenterology 136:1317-1327, e1311-1312.

Yamamoto Y, Yoshizawa T, Fukuda T, Shirode-Fukuda Y, Yu T, Sekine K, Sato T, Kawano H, Aihara K, Nakamichi Y, Watanabe T, Shindo M, Inoue K, Inoue E, Tsuji N, Hoshino M, Karsenty G, Metzger D, Chambon P, Kato S, and Imai Y (2013) Vitamin D receptor in osteoblasts is a negative regulator of bone mass control. Endocrinology 154:1008-1020.

Yamane T, Takeuchi K, Yamamoto Y, Li YH, Fujiwara M, Nishi K, Takahashi S, and Ohkubo I (2002) Legumain from bovine kidney: its purification, molecular cloning, immunohistochemical localization and degradation of annexin II and vitamin D-binding protein. Biochim Biophys Acta 1596:108-120.

Yasein N, Shroukh W, and Hijjawi R (2015) Serum vitamin D and the metabolic syndrome among osteoporotic postmenopausal female patients of a family practice clinic in Jordan. Adv Clin Exp Med 24:245-250.

Yin K, You Y, Swier V, Tang L, Radwan MM, Pandya AN, and Agrawal DK (2015) Vitamin D protects against atherosclerosis via regulation of cholesterol efflux and macrophage polarization in hypercholesterolemic swine. Arterioscler Thromb Vasc Biol 35:2432-2442.

Yu WD, McElwain MC, Modzelewski RA, Russell DM, Smith DC, Trump DL, and Johnson CS (1998) Enhancement of 1,25-dihydroxyvitamin D3-mediated antitumor activity with dexamethasone. J Natl Cancer Inst 90:134-141.

Zhang MC, Li HX, Liu HM, Lei H, Han L, Gao M, Mao JF, and Xu XJ (2014) Serum vitamin D is low and inversely associated with LDL cholesterol in the Kazak ethnic population: a cross- sectional study. Med Sci Monit 20:1274-1283.

Zhu L, Kong M, Han YP, Bai L, Zhang X, Chen Y, Zheng S, Yuan H, and Duan Z (2015) Spontaneous liver fibrosis induced by long term dietary vitamin D deficiency in adult mice is related to chronic inflammation and enhanced apoptosis. Can J Physiol Pharmacol 93:385-394.

Zierold C, Mings JA, and DeLuca HF (2001) Parathyroid hormone regulates 25-hydroxyvitamin D3- 24-hydroxylase mRNA by altering its stability. Proc Natl Acad Sci U S A 98:13572-13576.

Zierold C, Mings JA, and DeLuca HF (2003) Regulation of 25-hydroxyvitamin D3-24-hydroxylase mRNA by 1,25-dihydroxyvitamin D3 and parathyroid hormone. J Cell Biochem 88:234-237.

Zittermann A, Frisch S, Berthold HK, Gotting C, Kuhn J, Kleesiek K, Stehle P, Koertke H, and Koerfer R (2009) Vitamin D supplementation enhances the beneficial effects of weight loss on cardiovascular disease risk markers. Am J Clin Nutr 89:1321-1327.

Zollner G, Marschall HU, Wagner M, and Trauner M (2006) Role of nuclear receptors in the adaptive response to bile acids and cholestasis: pathogenetic and therapeutic considerations. Mol Pharmacol 3:231-251.

201

Appendix I – Supplementary material for Chapter 4

1,25(OH)2D3, atorvastatin, and combined 1,25(OH)2D3 and atorvastatin intervention study

We fed C57BL/6 mice a normal or high fat/high cholesterol (Western) diet for 3 weeks. Mice fed the normal diet were treated with vehicle (corn oil) i.p. daily. Mice fed the Western diet were treated with vehicle (corn oil) i.p. daily, 2.5 µg/kg 1,25(OH)2D3 i.p. q2d x 4 during the last week of Western diet, 10 mg/kg atorvastatin i.p. daily for the entire duration of Western diet, or a combination of

2.5 µg/kg 1,25(OH)2D3 i.p. q2d x 4 during the last week of Western diet and 10 mg/kg atorvastatin i.p. daily for the entire duration of Western diet.

Table A1-1. Body weights and plasma ALT at end of treatment period (mean ± SEM) Normal Diet Western Diet Vehicle- Vehicle- 1,25(OH) D 1,25(OH) D Atorvastatin 2 3 Treated Treated 2 3 + Atorvastatin Body Weight (g) 24.5 ± 0.8 25.2 ± 0.8 23.9 ± 0.3 26.0 ± 1.3 22.0 ± 1.0 Plasma ALT (IU/ml) 27.2 ± 4.3 17.0 ± 1.2 24.9 ± 1.9* 32.3 ± 2.4* 36.5 ± 2.2* # of mice per group 5 5 5 4 4 * P < 0.05 for Western diet vehicle-treated controls vs. 1,25(OH)2D3-, atorvastatin-, or combined treatment groups

202

Figure A1-1. Combined 1,25(OH)2D3 and atorvastatin treatment increases hepatic Cyp7a1 and decreases liver cholesterol in Western diet-fed mice. Combined 1,25(OH)2D3 and atorvastatin treatment (A) increased Cyp7a1 mRNA and protein expression, (B) decreased liver cholesterol concentrations, and (C) attenuated hepatic Shp and ileal Fgf15 mRNA levels in Western diet-fed mice (n = 4-5); P < 0.05 using the Student’s two-tailed t-test: †Western diet vs. normal diet; * Western diet-fed, vehicle-treated control vs. Western diet-fed, 1,25(OH)2D3-, atorvastatin-, or combined treated mice.

Figure A1-2. Changes in mRNA expression in the (A) liver, (B) ileum, and (C) kidney of normal and Western diet-fed mice treated with 1,25(OH)2D3, atorvastatin, or combined 1,25(OH)2D3 and atorvastatin. Data are mean ± SEM (n = 4-5). P < 0.05 using the Student’s two- tailed t-test: †Western diet vs. normal diet; * Western diet-fed, vehicle-treated control vs. Western diet-fed, 1,25(OH)2D3-, atorvastatin-, or combined treated mice.

203

Incubation of human liver slices with 1,25(OH)2D3

In collaboration with Dr. Geny M.M. Groothuis (University of Groningen, The Netherlands), we performed incubation studies in human precision-cut liver slices with 0.1% DMSO (control) or 100 nM 1,25(OH)2D3. The research protocols were approved by the Medical Ethical Committee of the University Medical Center (Groningen), with informed consent from the patients. Pieces of human liver tissue were obtained from patients undergoing partial hepactectomy for the removal of carcinoma or from redundant parts of donor livers remaining after split liver transplation as described by Elferink et al. (2011).

Figure A1-3. Incubation of human precision-cut liver slices in medium containing 0.1% DMSO or 100 nM 1,25(OH)2D3 for ATP assay and RNA or protein isolation.

204

Figure A1-4. (A) Effect of incubation on viability of human precision-cut liver slices (one donor, in triplicate) following 24 h incubation with 100 nM 1,25(OH)2D3, as measured by ATP content. (B) Relative mRNA expression in liver slices incubated for 0 or 24 h with 0.1% DMSO. At 0 h, mRNA expression of SHP, FGF19, LRH-1, CYP3A4 and CYP7A1 were significantly higher compared to samples treated with 0.1% DMSO for 24 h. Data are mean ± SEM for 0 h controls (n = 5, in triplicate); * P < 0.05 using the Student’s two-tailed t-test.

Figure A1-5. (A) Relative mRNA expression in human liver slices incubated for 24 h with 0.1% DMSO (white) or 100 nM 1,25(OH)2D3 (black). mRNA expression of CYP24A1 was significantly higher in 1,25(OH)2D3-treated slices, while CYP3A4 and CYP7A1 showed non- significant increases. Data are mean ± SEM for 1,25(OH)2D3-treated samples (n = 5, in triplicate); * P < 0.05 using the Student’s two-tailed t-test. (B) Optimized time (24 h) for CYP7A1 protein expression. Pooled liver slices from one donor were incubated for 0, 6, 16, or 24 h. (C) CYP7A1 protein expression in human liver slices. Protein expression was slightly higher in 1,25(OH)2D3- treated samples compared with control. Data are mean ± SEM (n = 5, in triplicate).

205

Appendix II – Supplementary material for Chapter 5

Table A2-1. Body weights (mean ± SEM) of mice at the end of the model validation study Vitamin D- Vitamin D- Vitamin D- Sufficient Deficient Deficient Weeks on Diet Control Control Control (0.47% Ca) (0.47% Ca) (2.5% Ca) (vehicle-treated) (vehicle-treated) (vehicle-treated) 0 27.1 ± 0.7 (4)a 2 25.7 ± 0.4 (3) 25.6 ± 0.6 (4) 27.2 ± 0.7 (6)

4 25.0 ± 0.7 (3) 30.4 ± 0.6† (5) 27.9 ± 1.1 (6) 6 26.9 ± 1.3 (3) 31.7 ± 1.4 (4) 31.2 ± 1.0 (7) 8 28.3 ± 0.3 (3) 29.2 ± 1.1 (4) 30.4 ± 0.8 (6) Mice used in Figs. 5-2 to 5-5; Supplementary Fig. S5-1 a Number of mice per group. † P < 0.05 for vitamin D-deficient vs. vitamin D-sufficient.

206

Table A2-2. Body weights (mean ± SEM) and plasma calcium and phosphorus measurements of mice at the end of the short-term 1,25(OH)2D3 intervention study without calcium supplementation Normal Diet High Fat/High Cholesterol Diet Vitamin D- Vitamin D- Vitamin D- Vitamin D- Vitamin D- Vitamin D- Sufficient Deficient Deficient + Sufficient Deficient Deficient + Control Control 1,25(OH)2D3 Control Control 1,25(OH)2D3 (vehicle-treated) (vehicle-treated) (vehicle-treated) (vehicle-treated) Body Weight (g) 30.2 ± 0.3 29.2 ± 1.1 29.4 ± 0.5 33.4 ± 1.0 32.7 ± 1.9 31.8 ± 0.8 Plasma Calcium 11.2 ± 0.3 11.4 ± 0.5 13.8 ± 0.2* 11.5 ± 0.5 11.8 ± 0.1 17.4 ± 1.5* (mg/dl) Plasma Phosphorus 19.4 ± 0.9 17.6 ± 1.0 18.2 ± 1.0 22.3 ± 1.5 22.5 ± 0.8 18.3 ± 1.1* (mg/dl) # of mice per group 3 4 6 4 4 5 Mice used in Fig. 5-7

Table A2-3. Body weights (mean ± SEM) and plasma calcium and phosphorus measurements of mice at the end of the short-term 1,25(OH)2D3 intervention study with calcium supplementation Normal Diet High Fat/High Cholesterol Diet Vitamin D- Vitamin D- Vitamin D- Vitamin D- Vitamin D- Vitamin D- Sufficient Deficient Deficient + Sufficient Deficient Deficient + Control Control 1,25(OH)2D3 Control Control 1,25(OH)2D3 (vehicle-treated) (vehicle-treated) (vehicle-treated) (vehicle-treated) Body Weight (g) 26.4 ± 0.3 30.4 ± 0.8† 26.3 ± 0.6* 31.3 ± 1.1# 33.4 ± 0.7 24.4 ± 0.7* Plasma Calcium 9.4 ± 0.4 8.9 ± 0.2 13.4 ± 0.6* 9.9 ± 0.4 10.8 ± 0.2 14.2 ± 0.7* (mg/dl) Plasma Phosphorus 17.0 ± 0.3 16.2 ± 0.8 14.5 ± 0.4* 17.7 ± 0.6 ± † 16.3 ± 0.8* (mg/dl) 20.9 0.9 # of mice per group 6 6 7 6 6 6 Mice used in Fig. A2-2 † P < 0.05 for vitamin D-deficient vs. vitamin D-sufficient vehicle-treated controls for respective diet groups. * P < 0.05 for vitamin D-deficient, vehicle-treated vs. 1,25(OH)2D3-treated for respective diet groups. # P < 0.05 for vitamin D-sufficient, normal vs. high fat/high cholesterol diet controls.

207

Table A2-4. Body weights (mean ± SEM) of mice at the end of the bile acid pool size study with short-term 1,25(OH)2D3 intervention without calcium supplementation Normal Diet High Fat/High Cholesterol Diet Vitamin D- Vitamin D- Vitamin D- Vitamin D- Vitamin D- Vitamin D- Sufficient Deficient Deficient + Sufficient Deficient Deficient + Control Control 1,25(OH)2D3 Control Control 1,25(OH)2D3 (vehicle-treated) (vehicle-treated) (vehicle-treated) (vehicle-treated) Body Weight (g) 29.2 ± 0.9 30.5 ± 1.1 28.1 ± 1.1 29.8 ± 2.2 32.6 ± 1.9 29.9 ± 1.1 # of mice per group 4 4 4 4 4 4 Mice used in Fig. 5-7C

Table A2-5. Body weights (mean ± SEM) of mice at the end of the bile acid pool size study with short-term 1,25(OH)2D3 intervention with calcium supplementation Normal Diet High Fat/High Cholesterol Diet Vitamin D- Vitamin D- Vitamin D- Vitamin D- Vitamin D- Vitamin D- Sufficient Deficient Deficient + Sufficient Deficient Deficient + Control Control 1,25(OH)2D3 Control Control 1,25(OH)2D3 (vehicle-treated) (vehicle-treated) (vehicle-treated) (vehicle-treated) Body Weight (g) 29.2 ± 0.9 28.8 ± 0.7 30.4 ± 1.0 29.8 ± 2.2 33.0 ± 0.6 32.0 ± 1.2 # of mice per group 4 4 4 4 4 4 Mice used in Fig. A2-2C

208

Table A2-6. Body weights (mean ± SEM) of mice at the end of the long-term vitamin D3 intervention study Normal Diet High Fat/High Cholesterol Diet Vitamin D- Vitamin D- Vitamin D- Vitamin D- Vitamin D- Vitamin D- Sufficient Deficient Deficient + Sufficient Deficient Deficient + Control Control Vitamin D3 Control Control Vitamin D3 (vehicle-treated) (vehicle-treated) (vehicle-treated) (vehicle-treated) Body Weight (g) 32.6 ± 2.7 34.2 ± 1.1 37.6 ± 2.0 35.6 ± 0.8 32.5 ± 4.1 35.2 ± 1.5 # of mice per group 4 5 6 4 5 6 Mice used in Fig. 5-8

Table A2-7. Body weights (mean ± SEM) of mice at the end of the bile acid pool size study with long-term vitamin D3 intervention Normal Diet High Fat/High Cholesterol Diet Vitamin D- Vitamin D- Vitamin D- Vitamin D- Vitamin D- Vitamin D- Sufficient Deficient Deficient + Sufficient Deficient Deficient + Control Control Vitamin D3 Control Control Vitamin D3 (vehicle-treated) (vehicle-treated) (vehicle-treated) (vehicle-treated) Body Weight (g) 35.9 ± 1.5 36.0 ± 1.9 35.4 ± 2.2 38.8 ± 1.0 35.6 ± 1.5 35.2 ± 1.8 # of mice per group 4 4 4 4 4 4 Mice used in Fig. 5-8C

209

Figure A2-1. Correlation of plasma 1,25(OH)2D3 vs. 25(OH)D3. Correlation of plasma 1,25(OH)2D3 with 25(OH)D3 in D-sufficient (white) mice and mice fed D-deficient diet without (grey) or with (black) Ca for 2, 4, 6 or 8 weeks revealed a hyperbolic function. Each datum point represents one individual mouse.

Figure A2-2. Short-term intervention with 1,25(OH)2D3 also decreases cholesterol in D- deficient diets supplemented with calcium. (A) Plasma and (B) liver cholesterol levels, which were elevated with the D-deficient diet (supplemented with Ca) were significantly reduced with 1,25(OH)2D3 treatment, whereas (C) total bile acid pool size (composed of t-CA, t-βMCA, t- αMCA, t-ωMCA, CA) was increased. (D) Treatment decreased Shp mRNA and increased Cyp7a1 mRNA and protein expression. Data are mean ± SEM (n = 3-6); P < 0.05 using the Student’s t- test: † D-sufficient vs. D-deficient controls; * D-deficient control vs. treated; # D-sufficient controls, normal vs. HF/HC.

210

Figure A2-3. Relative Srebp-2, Hmgcr, and Dhcr7 mRNA expression in vitamin D- deficiency. Vehicle-treated mice were fed D-sufficient (white) or D-deficient diets without (grey) or with (black) Ca supplementation for 8- or 11-weeks. The mRNA expression of Srebp-2 (A), Hmgcr (D), and Dhcr7 (G) in liver were increased slightly but non-significantly in mice after 8 weeks of the D-deficient diets. Data are mean ± SEM (n = 3-7); P < 0.05 using one-way ANOVA: † D-sufficient vs. D-deficient controls; # D-deficient controls without vs. with Ca supplementation. At the end of diets, no correlation was found between liver Srebp-2 (B), Hmgcr (E), and Dhcr7 (H) mRNA expression vs. liver 1,25(OH)2D3 levels. When liver cholesterol was plotted against Srebp-2 (C), Hmgcr (F), and Dhcr7 (I) mRNA expression, again, no correlations were found. Each datum point represents one individual mouse.

211

Figure A2-4. Treatment with 1,25(OH)2D3 and vitamin D3 do not alter expression of ileal Fgf15. Short-term intervention studies with 1,25(OH)2D3 in mice fed D-deficient diets for 8 weeks (A) without or (B) with Ca supplementation showed that treatment with 1,25(OH)2D3 did not alter expression of Fgf15. (C) Long-term intervention studies with vitamin D3 in mice fed D- deficient diet for 11 weeks followed a similar trend. Data are mean ± SEM (n = 3-6); P < 0.05 using the Student’s t-test: † D-sufficient vs. D-deficient controls; * D-deficient control vs. treated; # D-sufficient controls, normal vs. HF/HC.

212

Figure A2-5. Changes in hepatic mRNA expression of nuclear receptors, enzymes, and transporters in short-term intervention with 1,25(OH)2D3. Diet groups include: normal diet without calcium supplementation (ND –Ca), normal diet with calcium supplementation (ND +Ca), high fat/high cholesterol diet without calcium (HF/HC –Ca), and high fat/high cholesterol diet with calcium (HF/HC +Ca). Data are mean ± SEM (n = 3-6) and normalized to vitamin D-sufficient control in ND –Ca diet group; †P < 0.05 for D-sufficient vs. D-deficient controls in respective diet groups; *P < 0.05 for D-deficient control vs. treated in respective diet groups; #P < 0.05 for ND –Ca D-sufficient controls vs. other D-sufficient controls. 213

Figure A2-6. Changes in ileal mRNA expression of (A) nuclear receptors and (B) transporters in short-term intervention with 1,25(OH)2D3. Diet groups include: normal diet without calcium supplementation (ND –Ca), normal diet with calcium supplementation (ND +Ca), high fat/high cholesterol diet without calcium (HF/HC –Ca), and high fat/high cholesterol diet with calcium (HF/HC +Ca). Data are mean ± SEM (n = 3-6) and normalized to vitamin D-sufficient control in ND –Ca diet group; †P < 0.05 for D-sufficient vs. D-deficient controls in respective diet groups; *P < 0.05 for D-deficient control vs. treated in respective diet groups; #P < 0.05 for ND –Ca D-sufficient controls vs. other D-sufficient controls.

214

Figure A2-7. Changes in renal mRNA expression in short-term intervention with 1,25(OH)2D3. Diet groups include: normal diet without calcium supplementation (ND –Ca), normal diet with calcium supplementation (ND +Ca), high fat/high cholesterol diet without calcium (HF/HC –Ca), and high fat/high cholesterol diet with calcium (HF/HC +Ca). Data are mean ± SEM (n = 3-6) and normalized to vitamin D-sufficient control in ND –Ca diet group; †P < 0.05 for D-sufficient vs. D-deficient controls in respective diet groups; *P < 0.05 for D-deficient control vs. treated in respective diet groups; #P < 0.05 for ND –Ca D-sufficient controls vs. other D-sufficient controls.

215

Figure A2-8. Changes in hepatic mRNA expression in long-term intervention with D3. Diet groups include: normal diet without calcium supplementation (ND) and high fat/high cholesterol diet without calcium (HF/HC). Data are mean ± SEM (n = 4-6) and normalized to vitamin D-sufficient control in ND diet group; †P < 0.05 for D-sufficient vs. D-deficient controls; *P < 0.05 for D-deficient control vs. treated; #P < 0.05 for D-sufficient diet controls, normal vs. HF/HC.

216

Figure A2-9. Changes in ileal mRNA expression in long-term intervention with D3. Diet groups include: normal diet without calcium supplementation (ND) and high fat/high cholesterol diet without calcium (HF/HC). Data are mean ± SEM (n = 4-6) and normalized to vitamin D-sufficient control in ND diet group; †P < 0.05 for D-sufficient vs. D-deficient controls; *P < 0.05 for D-deficient control vs. treated; #P < 0.05 for D-sufficient diet controls, normal vs. HF/HC.

217

Figure A2-10. Changes in renal mRNA expression in long-term intervention with D3. Diet groups include: normal diet without calcium supplementation (ND) and high fat/high cholesterol diet without calcium (HF/HC). Data are mean ± SEM (n = 4-6) and normalized to vitamin D-sufficient control in ND diet group; †P < 0.05 for D-sufficient vs. D-deficient controls; *P < 0.05 for D-deficient control vs. treated; #P < 0.05 for D-sufficient diet controls, normal vs. HF/HC.

218

Appendix III – Supplementary material for Chapter 6

When compared with 1,25(OH)2D3, treatment with both 1α(OH)D3 and lithocholic acid acetate (LCAa), an alternate ligand of the VDR, produced similar effects on mRNA expression of VDR,

CYP24, CYP3A4, TRPV6, and OATP1A2. Meanwhile, treatment with 1α(OH)D3 also induced

MDR1 and MRP2 to similar levels as 1,25(OH)2D3 treatment. Unlike 1,25(OH)2D3, treatment with LCAa downregulated MDR1 and increased ASBT mRNA expression.

Figure A3-1. Changes in mRNA expression of nuclear receptors, transporters, and enzymes in Caco-2 cells grown for 21 days and treated on day 18 for 3 days with 0.1% DMSO, 100 nM 1,25(OH)2D3, 100 nM 1α(OH)D3, or 10 µM LCAa. Data are mean ± SD (n = 3, in triplicate); *P < 0.05 for DMSO control vs. treated.

Since vitamin D analogs are prone to triggering hypercalcemia, we also investigated the cholesterol lowering effects of alternate ligands of the VDR, namely, LCA derivatives. In vivo studies were performed on C57BL/6 mice treated i.p. with 0, 0.5, or 0.75 mmol/kg of LCAa or LCAaMe (lithocholic acid acetate methyl ester) q2d x 4. Both LCA derivatives appeared to lack significant effects on cholesterol lowering, although the 0.75 mmol/kg treated groups displayed significant induction of Cyp7a1 protein expression.

219

Figure A3-2. Changes in hepatic (A) mRNA and (B) Cyp7a1 protein expression and (C) plasma cholesterol levels and mRNA expression in (D) ileum and (E) kidney following treatment with LCAa. Data are mean ± SEM (n = 4-5); *P < 0.05 for 0 mmol/kg vs. treated.

Figure A3-3. Changes in hepatic (A) mRNA and (B) Cyp7a1 protein expression following treatment with LCAaMe. Data are mean ± SEM (n = 4-5); *P < 0.05 for 0 mmol/kg vs. treated.

220

In addition to the use of vitamin D analogs or alternate ligands of the VDR to avoid hypercalcemia, we also investigated the effects of a protracted long-term treatment regimen with 1,25(OH)2D3. rd Here, mice were treated i.p. with vehicle (corn oil) or 2.5 µg/kg 1,25(OH)2D3 every 3 day for 8 weeks. The 1,25(OH)2D3-treated mice had non-significantly reduced mRNA expression of Shp and significantly increased mRNA expression of Cyp7a1, without triggering any hypercalcemia (Durk et al., 2014).

Figure A3-4. Changes in hepatic mRNA expression following 1,25(OH)2D3 treatment q3d x 19. Data are mean ± SEM (n = 4-6); *P < 0.05 for control vs. treated.

221

Appendix IV – Supplementary material for Chapter 7

Table A4-1. Body weights (mean ± SEM) for mice at the end of the single i.v. dose study First Dose Time (h) 2 pmol 60 pmol 120 pmol 0 21.7 ± 0.3 (3)a 3 22.7 ± 0.7 (4) 22.5 ± 0.4 (4) 22.0 ± 0.9 (4) 6 21.4 ± 0.4 (4) 22.0 ± 0.5 (4) 21.4 ± 0.3 (4) 9 19.7 ± 0.4* (4) 21.8 ± 0.7 (4) 22.5 ± 0.4 (4) 12 22.7 ± 0.3 (4) 21.8 ± 0.6 (4) 22.5 ± 0.6 (4) 24 21.9 ± 0.6 (4) 22.4 ± 0.3 (4) 21.9 ± 0.7 (4) 48 23.0 ± 0.3* (4) 22.1 ± 0.6 (4) 22.7 ± 0.4 (4) Mice used in Figs. 7-4 to 7-8. a Number of mice per group. * P < 0.05 vs. 0 h control of respective treatment group.

Table A4-2. Body weights (mean ± SEM) for mice at the end of the repeated i.v. dose study Repeated Third Dose Time (h) 2 pmol 120 pmol 0 24.7 ± 0.8 (4)a 27.2 ± 0.5 (5) 3 25.3 ± 0.4 (4) 25.6 ± 0.9 (4) 6 24.6 ± 0.5 (4) 24.8 ± 1.1 (4) 9 24.5 ± 0.7 (4) 24.6 ± 0.5* (3) 12 25.3 ± 1.1 (4) 26.1 ± 0.6 (4) 24 25.2 ± 0.6 (4) 24.1 ± 0.7* (4) 48 24.1 ± 0.7 (4) 25.7 ± 0.4 (4) Mice used in Figs. 7-4 to 7-8. a Number of mice per group. * P < 0.05 vs. 0 h control of respective treatment group.

222

Figure A4-1. Relative mRNA expression of (A) renal and (B) ileal Vdr following single or repeated i.v. administration of 1,25(OH)2D3. Data for vehicle-treated mice (basal level) were averaged and joined by the dashed line (n = 9), whereas data for treated mice are mean ± SEM and joined by a solid line (n = 3-4 different mice). Significant differences between groups were denoted by: a, basal level versus 2 pmol; b, basal level versus 60 pmol; c, basal level versus 120 pmol; †, 2 pmol versus 60 pmol; ‡, 2 pmol versus 120 pmol; #, 60 pmol versus 120 pmol; *, all groups except basal level versus 2 pmol.

223

Figure A4-2. Changes in (A) plasma calcium and (B) renal and (C) ileal Trpv6 mRNA expression following single or repeated i.v. administration of 1,25(OH)2D3. Data for vehicle- treated mice (basal level) were averaged and joined by the dashed line (n = 9), whereas data for treated mice are mean ± SEM and joined by a solid line (n = 3-4 different mice). Significant differences between groups were denoted by: a, basal level versus 2 pmol; b, basal level versus 60 pmol; c, basal level versus 120 pmol; †, 2 pmol versus 60 pmol; ‡, 2 pmol versus 120 pmol; #, 60 pmol versus 120 pmol; *, all groups except basal level versus 2 pmol.

224

Gastroenterology 2014;146:1048–1059 BASIC AND TRANSLATIONAL—LIVER Vitamin D Receptor Activation Down-regulates the Small Heterodimer Partner and Increases CYP7A1 to Lower Cholesterol Edwin C. Y. Chow,1 Lilia Magomedova,1 Holly P. Quach,1,* Rucha Patel,1,* Matthew R. Durk,1 Jianghong Fan,1 Han-Joo Maeng,1 Kamdi Irondi,1 Sayeepriyadarshini Anakk,2 David D. Moore,2 Carolyn L. Cummins,1,§ and K. Sandy Pang1,§

1Department of Pharmaceutical Sciences, Leslie Dan Faculty of Pharmacy, University of Toronto, Toronto, Ontario, Canada and 2Department of Molecular and Cellular Biology, Baylor College of Medicine, Houston, Texas

Keywords: Liver; Bile Acid; Farnesoid X Receptor; Transcrip- See editorial on page 899. tional Regulation.

BACKGROUND & AIMS: Little is known about the effects of the holesterol is an essential component of cell mem- vitamin D receptor (VDR) on hepatic activity of human branes and the precursor to steroid hormones and cholesterol 7a-hydroxylase (CYP7A1) and cholesterol meta- C bolism. We studied these processes in mice in vivo and mouse bile acids. In excess, cholesterol can lead to atherosclerosis and human hepatocytes. METHODS: Farnesoid X receptor and coronary heart disease. In liver, cholesterol is metabo- a (Fxr) / , small heterodimer partner (Shp) / , and C57BL/6 lized to bile acids by cholesterol 7 -hydroxylase (CYP7A1), (wild-type control) mice were fed normal or Western diets for which is the rate-limiting metabolic enzyme in the classic 1 3 weeks and were then given intraperitoneal injections of bile acid synthetic pathway. The CYP7A1 promoter contains vehicle (corn oil) or 1a,25-dihydroxyvitamin D3 (1,25[OH]2D3; highly conserved bile acid responsive regions known to be 4doses,2.5mg/kg, every other day). Plasma and tissue sam- modulated by feedback repression by various transcription ples were collected and levels of Vdr, Shp, Cyp7a1, Cyp24a1, factors in response to increasing hepatic bile acid concen- and rodent fibroblast growth factor (Fgf) 15 expression, as trations.1 A primary, negative feedback mechanism of well as levels of cholesterol, were measured. We studied the CYP7A1 regulation is the human farnesoid X receptor ([FXR] regulation of Shp by Vdr using reporter and mobility shift NR1H4) and human small heterodimer partner ([SHP] assays in transfected human embryonic kidney 293 cells, NR0B2) regulatory cascade.2 Bile acids such as chenodeoxy- RNLTOA LIVER TRANSLATIONAL quantitative polymerase chain reaction with mouse tissues cholic acid (CDCA) activate FXR to increase transcription of AI AND BASIC and mouse and human hepatocytes, and chromatin immuno- SHP, an atypical nuclear receptor that lacks a DNA binding precipitation assays with mouse liver. RESULTS: We first domain and represses CYP7A1 activation by suppression fi con rmed the presence of Vdr mRNA and protein expression of transcription factors, liver receptor homolog-1 (NR5A2) in livers of mice. In mice fed normal diets and given injections and hepatocyte nuclear factor 4a (NR2A1), which are of 1,25(OH)2D3, liver and plasma concentrations of 1,25 essential for CYP7A1 expression.3 A second negative feed- (OH)2D3 increased and decreased in unison. Changes in he- back mechanism on CYP7A1 is found in the intestine, where patic Cyp7a1 messenger RNA (mRNA) correlated with those of activation of FXR induces fibroblast growth factor 15/19 Cyp24a1 (a Vdr target gene) and inversely with Shp mRNA, but (rodent/human), a hormonal signaling molecule that re- not ileal Fgf15 mRNA. Similarly, incubation with 1,25(OH) D 2 3 presses CYP7A1 through interaction with the liver fibroblast increased levels of Cyp24a1/CYP24A1 and Cyp7a1/CYP7A1 4 mRNA in mouse and human hepatocytes, and reduced levels growth factor receptor 4 via the c-Jun signaling pathway. of Shp mRNA in mouse hepatocytes. In Fxr/ and wild-type mice with hypercholesterolemia, injection of 1,25(OH)2D3 consistently reduced levels of plasma and liver cholesterol and *Authors share co-third authorship; §Authors share co-senior authorship. Shp mRNA, and increased hepatic Cyp7a1 mRNA and protein; these changes were not observed in Shp / mice given Abbreviations used in this paper: Asbt, rodent apical sodium dependent bile acid transporter; CDCA, chenodeoxycholic acid; ChIP, chromatin 1,25(OH)2D3 andfedWesterndiets.Truncationofthehuman immunoprecipitation; Cyp7a1/CYP7A1, rodent/human cholesterol 7a- small heterodimer partner (SHP) promoter and deletion ana- hydroxylase; 1,25[OH]2D3,1a,25-dihydroxyvitamin D3; EMSA, electro- lyses revealed VDR-dependent inhibition of SHP,andmobility mobility shift assays; Fgf15, rodent fibroblast growth factor 15; Fxr/ FXR, rodent/human farnesoid X receptor; HDL-C, high-density lipo- shift assays showed direct binding of VDR to enhancer regions protein cholesterol; HMGCo-A, human 3-hydroxy-3-methyl-glutaryl- of SHP. In addition, chromatin immunoprecipitation analysis CoA; LDL-C, low-density lipoprotein cholesterol; Lrh-1, rodent liver of livers from mice showed that injection of 1,25(OH) D receptor homolog 1; mRNA, messenger RNA; Rxra, rodent retinoid X 2 3 receptor-a; Shp/SHP, rodent/human small heterodimer partner; VDRE, increased recruitment of Vdr and rodent retinoid X receptor to vitamin D receptor response element. the Shp promoter. CONCLUSIONS: Activation of the VDR represses hepatic SHP to increase levels of mouse and human © 2014 by the AGA Institute CYP7A1 and reduce cholesterol. 0016-5085/$36.00 http://dx.doi.org/10.1053/j.gastro.2013.12.027 April 2014 VDR Up-Regulates Cyp7a1 and Lowers Cholesterol 1049

The vitamin D receptor ([VDR] NR1I1) binds to its an indirect mechanism.15 The divergent views are partially endogenous ligand, 1a,25-dihydroxyvitamin D3 (1,25 reconciled by species differences, with VDR protein levels 5 [OH]2D3) or lithocholic acid (alternate VDR ligand) to being extremely low in rat liver, but present at detectable activate the transcription of genes. Various mechanisms levels in mouse and man.16 The down-regulation of Cyp7a1 6 implicate a role for VDR on CYP7A1 regulation. Activation after 1,25(OH)2D3 treatment in the rat is explained as a of VDR was found to antagonize the CDCA-dependent secondary hepatic Fxr and not Vdr effect, because increased transactivation of FXR in VDR-transfected HepG2 cells7 bile acid absorption into portal blood occurred as a result of and blunt the Lxra-mediated induction of Cyp7a1 mRNA Vdr-mediated induction of the intestinal apical sodium- in rat hepatoma cells.8 VDR inhibition of CYP7A1 tran- dependent bile acid transporter (Asbt).15 Clinical reports scription in human hepatocytes and HepG2 cells has been relating vitamin D status to cholesterol (cholesterol levels vs attributed to blockage of hepatocyte nuclear factor 1,25[OH]2D3 or its less active precursor, 25-hydroxyvitamin 9 17 4amediated activation of CYP7A1. Induction of intestinal D3, 25[OH]D3) are equivocal. Both short-term and long- 18 Fgf15 after a high dose of 1,25(OH)2D3 in mice was shown term vitamin D treatment did not improve lipid proﬁles or to down-regulate Cyp7a1 messenger RNA (mRNA) level.10 lower cholesterol and only resulted in slightly lower serum By contrast, Vdr-knockout mice are reported to have triglyceride, while other studies documented increased higher total serum cholesterol,11 and treatment with 1a- high-density lipoprotein cholesterol (HDL-C),19 or hydroxyvitamin D3, a potent 1,25(OH)2D3 precursor, up- decreased total cholesterol and triglyceride but unchanged regulated Cyp7a1 mRNA expression in mice.12,13 Likewise, HDL-C and low-density lipoprotein cholesterol (LDL-C) doxercalciferol, a vitamin D analogue, decreased the accu- levels.20 A recent, population-based study established an mulation of triglycerides and cholesterol in murine kid- association between 1,25(OH)2D3 and HDL-C and 25(OH)D3 14 21 ney. In rat, 1,25(OH)2D3 down-regulated liver Cyp7a1 by and total cholesterol, LDL-C, and triglyceride. Atorvastatin, BASIC AND TRANSLATIONAL LIVER

Figure 1. Vdr protein distribution in mouse and liver immunostaining. (A) Nuclear Vdr protein expression in 50 mg ileum (I), liver (L), kidney (K), and brain (Br) of wild-type and Fxr/ mice, with human embryonic kidney 293 cells transfected with CMX (basal levels) () or overexpressing CMX-Vdr (þ) as standards (top panel), vs kidney samples of Vdr/ and Vdrþ/þ mice (lower panel). Vdr protein expression was similar between Fxrþ/þ and Fxr/ livers (approximately 30% of I or K), but absent in Vdr/ kidney (n ¼ 34); *,#P < .05 vs ileum for Fxrþ/þ and Fxr/ mice, respectively (Mann-Whitney U test). (B) Vdr protein (arrows) was present within nuclei of hepatocytes of wild-type mice. 1050 Chow et al Gastroenterology Vol. 146, No. 4

supplemented with vitamin D, further lowered total role of the VDR in liver cholesterol regulation remains cholesterol and LDL-C,22 and patients receiving statin or controversial. niacin supplemented with vitamin D and ﬁsh oil showed The intent of this study was to clarify the impact of Vdr on reduced LDL-C and triglycerides and elevated HDL-C.23 The Cyp7a1 regulation and cholesterol lowering. We veriﬁed the RNLTOA LIVER TRANSLATIONAL AI AND BASIC April 2014 VDR Up-Regulates Cyp7a1 and Lowers Cholesterol 1051

presence of Vdr in murine liver, and demonstrated that Western diet (Harlan Teklad Cat #88137; high-fat [42%]/high- 1,25(OH)2D3 given to mice rapidly reached the liver, result- cholesterol [0.2%] diet) for 3 weeks were established, a period ing in induction of hepatic Cyp7a1 via down-regulation of during which Cyp7a1 mRNA/protein expression was un- rodent small heterodimer partner (Shp). We showed that changed. These mice were given 4 repeated doses of Vdr-mediated Shp repression and up-regulation of Cyp7a1 1,25(OH)2D3 every other day at the beginning of the third week was Fxr independent but Shp dependent in vivo. Luciferase of the Western diet. On day 8, systemic and portal blood reporter, electrophoretic mobility shift assay (EMSA), and samples and tissues were obtained under anesthesia, as previously described.15,24 Basal mRNA levels of intestinal and liver chromatin immunoprecipitation (ChIP) assays further sup- / / ported a direct role for Vdr/VDR in the repression of Shp/ genes in Fxr and Shp mice were compared with those of SHP to result in up-regulation of Cyp7a1/CYP7A1, providing wild-type mice (Supplementary Figure 1A and B). 1,25(OH)2D3 treatment resulted in a slight elevation of plasma calcium, but a novel mechanism for cholesterol lowering. In mouse relatively unchanged phosphorous, alanine aminotransferase, and human hepatocytes, Cyp7a1/CYP7A1 expression was and portal bile acid levels (Supplementary Table 1). elevated on exposure to 1,25(OH)2D3. Mouse Primary Hepatocytes Materials and Methods Mouse primary hepatocytes were isolated26 and treated In the Supplementary Material, we provide information on with vehicle (0.1% EtOH) or 100 nM 1,25(OH)2D3 (M199 media materials, mouse strains, antibodies, plasmids, and procedures without fetal bovine serum) for 0, 3, 6, 9, 12, and 24 hours. Cells for real-time polymerase chain reaction, Western blotting, were harvested for mRNA determination at 9 hours, the mini- microsomal preparation for Cyp7a1 activity, and assay pro- mal time required for induction of Cyp24a1.27 Dose-dependent cedures for bile acid pool size, cholesterol, and 1,25(OH)2D3. activation of Cyp24a1 by 1,25(OH)2D3 (10 to 250 nM) was observed (data not shown). Immunostaining of Murine Liver Vdr Livers of male C57BL/6 mice were perfusion-fixed with Human Hepatocytes phosphate-buffered saline and 4% paraformaldehyde, and kept Upon arrival, the cold preservation medium for storage of overnight at 4C. Then, 7-mm-thick paraffin-embedded sections human primary hepatocytes (gift from Dr Jasminder Sahi, Life were dewaxed and incubated in 2N HCl at 37C for 30 minutes. Technologies, TX) was replaced with Williams E medium (Cat# Sections were preblocked with 5% goat serum in phosphate- CM4000, Life Technologies, TX), supplemented with the cell buffered saline containing 0.1% Tween-20, then incubated maintenance cocktail (10 mM of dexamethasone, 120 U/mL with the primary anti-VDR antibody 9A7 (1:50 v/v) overnight. penicillin, and 120 U/mL streptomycin, 2 mM GlutaMax, 15 mM After rinsing 3 times with 5% goat serum in phosphate- of HEPES and ITSþ 6.25 g/mL insulin, 6.25 g/mL transferrin, buffered saline containing 0.1% Tween-20, the secondary 6.25 g/mL selenous acid, 1.25 mg/mL bovine serum albumin, goat anti-rat horseradish peroxidase antibody was added for and 5.33 g/mL linoleic acid were used in dilutions according to 6 2 hours at room temperature and visualized using a metal- manufacture’s instructions). Plates (24-wells; 2.1 10 2 enhanced 3,30-diaminobenzidine tetrahydrochloride substrate cells/cm ) were acclimatized overnight at 37 C in a humidified kit (Thermo Scientific, Rockford, IL). After several washes, incubator. On the next day, cells were treated with vehicle (0.1% ethanol) or 100 nM 1,25(OH)2D3 and harvested at 3, 6, sections were imaged using a Nikon E1000R microscope. BASIC AND 12, and 24 hours. Activation of VDR was confirmed by CYP24A1

mRNA induction. TRANSLATIONAL LIVER 1,25(OH)2D3 Treatment of Mice in vivo In each set of in vivo studies, doses of 0 or 2.5 mg/kg Transfection Assays of Human Embryonic 1,25(OH)2D3, in sterile corn oil were given intraperitoneally every other day for 8 days at 9 to 10 AM.24 First, we treated Kidney 293 Cells normal-dietfed, male C57BL/6 (wild-type or Fxrþ/þ] and Cell transfection was performed in media containing 10% / Fxr mice (8 to 12 weeks; n ¼ 410) with 1,25(OH)2D3, and charcoal-stripped fetal bovine serum using calcium phosphate blood and tissues were harvested on day 8 between 12 PM and in 96-well plates. The total amount of plasmid DNA (150 ng/ 2 PM. In the second study, blood samples and livers from wild- well) included 50 ng reporter, 20 ng pCMX-b-galactosidase, type mice were collected at different time points during the 15 ng nuclear receptor, 15 ng pCMX-liver receptor homolog 1 25 1,25(OH)2D3 treatment period, as described. In the third (Lrh-1), and pGEM ﬁller plasmid. Ligands were added at 6 to 8 study, hypercholesterolemic models composed of wild-type, hours post transfection. Cells harvested 14 to 16 hours later Fxr / , and Shp / mice (68 weeks old; n ¼ 410) fed a were assayed for luciferase and b-galactosidase activity.

Figure 2. 1,25(OH)2D3 treatment increases Cyp7a1 mRNA, protein, and microsomal activity and decreases hepatic Shp expression in normal-dietfed wild-type and Fxr/ mice. (A) Cyp7a1 mRNA, protein (normalized to rodent glyceraldehyde-3- / phosphate dehydrogenase), and microsomal activity were increased after 1,25(OH)2D3 treatment in both wild-type and Fxr mice. (B) Hepatic Shp mRNA was decreased, and ileal (C) Fgf15 and (D) Asbt mRNA were unchanged (n ¼ 410); #P < .05 between Fxr/ vs wild-type vehicle-treated mice; *P < .05 between vehicle vs treated mice of the same genotype (Mann- Whitney U test). A signiﬁcant, negative correlation (each point represents one mouse) exists between (E) Cyp7a1 mRNA/ protein/activity and liver Shp (F), but not ileal Fgf15 mRNA in wild-type mice (n ¼ 34). 1052 Chow et al Gastroenterology Vol. 146, No. 4

Luciferase values were normalized to b-galactosidase to control was added and incubated for 30 minutes on ice. Reactions were for transfection efﬁciency and expressed as relative luciferase loaded onto a 6% polyacrylamide gel, transferred to Biodyne B units. membrane (Pierce) and cross-linked. Detection was carried out using the Light Shift Chemiluminescent EMSA Kit (Pierce). Nuclear Protein Extracts Human embryonic kidney 293 cells were transfected with ChIP 10 mg Vdr, rodent retinoid X receptor-a (Rxra), or CMX ChIP assays were performed on frozen livers after vehicle or plasmid DNA using calcium phosphate in 10-cm plates. At 30 1,25(OH)2D3 treatment in normal-diet fed wild-type mice (3 4 hours post transfection, nuclear protein extracts were sets of experiment, in triplicate) at 12 hours after the fourth 26 prepared.26 injection, as described. Liver nuclei were resuspended in 2 pellet volume of sonication buffer (0.2% sodium dodecyl sulfate, 1% Triton-X, 0.1% Na-deoxycholate, 1 mM EDTA and 50 mM Electrophoretic Mobility Shift Assay Tris-HCl [pH 8.0], 150 mM NaCl, and protease inhibitors). Oligonucleotide sequences are described in Supplementary Chromatin, diluted 2-fold in buffer (sonication buffer without Table 2. Oligos were biotinylated using a biotin 30 end DNA sodium dodecyl sulfate) and 400 mL diluted sample, were incu- labeling kit (Pierce, Rockford, IL). DNA binding reactions con- bated with 10 mg Vdr, Rxra, Lrh-1, or 2 mg H3K9me3 antibodies sisted of 2.5 mL of NE-PER nuclear extracts in binding buffer (see Supplementary Material) overnight and used for immuno- (10 mM Tris, 50 mM KCl, 1 mM dithiothreitol, 1 mM EDTA, 5% precipitation. After spin column puriﬁcation, the eluate was glycerol, 1 mg BSA and 1 mg poly(dI-dC); [pH 7.5]). After a diluted 5-fold with water, and quantitative polymerase chain 20-minute preincubation, 50 fmol annealed biotinylated DNA reaction was performed using 5 mL template DNA with the RNLTOA LIVER TRANSLATIONAL AI AND BASIC

Figure 3. Correlation between liver 1,25(OH)2D3 concentration and hepatic Cyp24a1 and Cyp7a1 mRNA expression in normal- dietfed wild-type mice. Repeated administration of 1,25(OH)2D3 resulted in (A) biphasic decay of 1,25(OH)2D3 concentration in liver (mean points; n ¼ 24; solid and open circles for levels in treated livers or basal levels in vehicle-treated livers) that paralleled those in plasma (gray solid and open symbols for treated and basal levels in vehicle-treated livers; previously published25) and corresponding changes in hepatic (B) Cyp24a1 mRNA, (C) Cyp7a1 mRNA, (D) Cyp7a1 protein expression (n ¼ 24). The insets show the diurnal variation of basal Cyp7a1 mRNA and protein expression, peaking at around 9 PM and 12 AM, respectively, in vehicle-treated mice. For (B)to(D), each point represents datum from one mouse, except for insets in (C) and (D), where the open symbols denote mean values (n ¼ 24). April 2014 VDR Up-Regulates Cyp7a1 and Lowers Cholesterol 1053 following primer set: forward: 50-GAGCGCCTGAGACCTTGGT-30 Vdr protein was identified in mouse hepatocytes by im- and reverse: 50-TCAAGTGCATAAACAGGGTCATTAA-30 ampli- munostaining (Figure 1B), and specificity of the antibody fying the putative mouse Shp VDRE. Quantification was per- was further confirmed by staining liver sections or West- formed by quantitative polymerase chain reaction (standard ern blots of livers obtained from Vdr/ mice (data not curve method) using serial dilutions of the input as standards. shown).

Statistics 1,25(OH)2D3 Increases Hepatic Cyp7a1 and Data are expressed as mean SEM for in vivo data and Decreases Hepatic Shp But Not Ileal Fgf15 in mean SD for in vitro data. For comparison of in vivo and þ/þ / in vitro data between 2 groups, the Mann-Whitney U test Normal-DietFed Fxr and Fxr Mice / and the unpaired Student t tests were used, respectively, and The Fxr mouse was used to circumvent potential P < .05 was set as the level of significance. confounding effects of feedback regulation of cholesterol metabolism through hepatic Fxr.15 Absence of Fxr in Fxr / mice resulted in higher basal hepatic Cyp7a1 mRNA, protein, Results and microsomal activity (Figure 2A) compared to those of Fxrþ/þ mouse, and lower mRNA basal levels of hepatic Shp Vdr Protein Tissue Distribution and and intestinal Fgf15 and Asbt (Figure 2BD). Remarkably, Liver Immunostaining 1,25(OH)2D3 treatment resulted in significant up-regulation Nuclear Vdr protein was present at similar levels in the of hepatic Cyp7a1 mRNA and protein expression and ileum and kidney of wild-type and Fxr/ mice, although microsomal activity in both Fxrþ/þ and Fxr/ genotypes levels were considerably lower in liver and brain, as found (Figure 2A), accompanied by a reduction in hepatic Shp previously.24,25 Vdr protein was found in the lysate of mRNA expression without changes in intestinal Fgf15 primary hepatocytes prepared from wild-type mice (data and Asbt (Figures 2BD). There was a significant, negative þ þ not shown) and Vdr / but not Vdr / kidney (Figure 1A). correlation between Cyp7a1 mRNA/protein/microsomal BASIC AND TRANSLATIONAL LIVER

Figure 4. 1,25(OH)2D3 treatment changes Cyp24a1/CYP24A1 and Cyp7a1/CYP7A1, and Shp mRNA expression in wild-type mouse and human primary hepatocytes. (A) Freshly isolated mouse hepatocytes showed increased Cyp7a1 and Cyp24a1 and reduced Shp mRNA expression at 9 hours after 100 nM 1,25(OH)2D3 treatment and (B) human hepatocytes (from the same donor) exposed to 100 nM 1,25(OH)2D3 vs vehicle showed time-dependent induction of CYP24A1 and CYP7A1 mRNA and CYP7A1 protein. Data are from 1 donor and presented as mean SD (n ¼ 3) of triplicates. P < .05: *compared with vehicle control (t test). 1054 Chow et al Gastroenterology Vol. 146, No. 4

activity and hepatic Shp (Figure 2E), but not intestinal Fgf15 hours post injection, and levels were amplified with sub- mRNA expression (Figure 2F), suggesting that attenuation of sequent injections (60 and 80-fold higher; Figure 3C). Pat- Shp resulted in elevated Cyp7a1, independent of Fxr. terns of Cyp7a1 mRNA and protein induction were similar (Figures 3C and 3D). The increase in Cyp7a1 mRNA expression in response to 1,25(OH)2D3 was much higher Parallel Changes in Liver 1,25(OH)2D3 (60- to 80-fold) than the peak of the circadian rhythm (6.7- Concentrations and Temporal Cyp7a1 fold, Figure 3C). and Cyp24a1 mRNA Expression in Wild-Type Mice 1,25(OH) D Increases Cyp7a1/CYP7A1 and To further examine Cyp7a1 expression changes in 2 3 response to Vdr activation, we systematically analyzed the Cyp24a1/CYP24A1 mRNA Levels in Mouse relationship between liver 1,25(OH)2D3 concentration and and Human Primary Hepatocytes gene expression in normal-dietfed wild-type mice To confirm that changes in Cyp7a1 and Shp mRNA were throughout the 1,25(OH)2D3 treatment period. A diurnal independent of other physiologic signaling molecules from variation was found in both basal Cyp7a1 mRNA and pro- the gut or portal circulation, isolated mouse primary hepa- tein levels (insets of Figure 3C and D), with peaks occurring tocytes were incubated with 100 nM 1,25(OH)2D3. Asignifi- at around 9 PM and 12 AM, respectively. After 1,25(OH)2D3 cant decrease in Shp (35%) with subsequent increase in dosing, a biphasic decay profile of liver 1,25(OH)2D3 that Cyp7a1 (3-fold) and Cyp24a1 (19-fold) mRNA expression closely paralleled the plasma concentration-time curve25 was observed at 9 hours (Figure 4A), confirming an auton- was observed (Figure 3A). In response to increased liver omous role for 1,25(OH)2D3 in the modulation of Cyp7a1 via 1,25(OH)2D3, hepatic Cyp24a1 mRNA levels rose, occurring SHP repression in hepatocytes. Human primary hepatocytes maximally between 3 and 6 hours post injection (Figure 3B). exposed to 100 nM 1,25(OH)2D3 showed increased human Cyp7a1 mRNA expression also rose maximally at around 12 CYP24A1 and CYP7A1 mRNA expression at 12 and 24 hours, RNLTOA LIVER TRANSLATIONAL AI AND BASIC

Figure 5. 1,25(OH)2D3 treatment increases hepatic Cyp7a1 and decreases hepatic Shp mRNA expression and plasma and liver cholesterol in Western-dietfed, wild-type mice. (A) Hepatic Cyp7a1 protein and microsomal activity were increased after 1,25(OH)2D3 treatment. (B) Plasma and liver cholesterol concentrations, elevated with the Western diet, were reduced after 1,25(OH)2D3 treatment compared with Western-dietfed controls. (C) Hepatic Shp mRNA expression was increased with Western diet, which decreased with 1,25(OH)2D3 treatment. The Western diet increased ileal Fgf15 mRNA level, which † remained relatively unchanged with 1,25(OH)2D3 treatment (n ¼ 48); P < .05: Western diet vs normal diet; *Western- dietfed, vehicle-treated control vs Western-dietfed, 1,25(OH)2D3-treated mice (Mann-Whitney U test). April 2014 VDR Up-Regulates Cyp7a1 and Lowers Cholesterol 1055 respectively, and CYP7A1 protein expression at 24 hours in other cholesterol-related genes in the intestine and liver (Figure 4B). Hepatocytes from 2 other human donors dis- (Supplementary Figure 2). played similar trends (data not shown). In Hypercholesterolemic Fxr/ and Shp/ In Hypercholesterolemic Wild-Type Mice, Mouse Models, 1,25(OH)2D3 Reduced / 1,25(OH)2D3 Increases Hepatic Cyp7a1 and Plasma and Liver Cholesterol in Fxr Lowers Cholesterol by Decreasing Hepatic Mice But Not Shp/ Mice Shp Without Changing Ileal Fgf15 To examine whether Fxr and Shp are involved in The Western diet did not alter Cyp7a1 expression or cholesterol lowering, we fed Fxr/ and Shp/ mice with microsomal activity of wild-type mice (Figure 5A), but the same Western diet and used the same 1,25(OH)2D3 in- increased plasma and liver cholesterol levels (Figure 5B) jection regimen as that for wild-type mice. Western and hepatic Shp and ileal Fgf15 mRNA expression diet alone did not alter basal Cyp7a1 levels (Figure 6A and (Figure 5C). The 1,25(OH)2D3 treatment increased Cyp7a1 D), but increased plasma and liver cholesterol concentra- protein expression (76%) and microsomal activity (280%) tions in Western-dietfed Fxr / (Figure 6B), although not (Figure 5A) and lowered both plasma and liver cholesterol Western-dietfed Shp/ mice (Figure 6E). Basal hepatic (Figure 5B) and hepatic Shp mRNA expression (Figure 5C). Shp mRNA level was higher, and ileal Asbt mRNA The bile acid pool size and fecal bile acid excretion were level, slightly lower, in Western-dietfed compared with / signiﬁcantly increased after 1,25(OH)2D3 treatment normal-dietfed Fxr controls (Figure 6C), and basal ileal (Supplementary Figure 1C), suggesting that increased Fgf15 mRNA expression was higher in both Western- Cyp7a1 activity led to more bile acid formation. There was, dietfed Fxr/ and Shp/ controls (Figures 6C and F). however, little change in the portal bile acid concentration The 1,25(OH)2D3 treatment increased Cyp7a1 mRNA (Supplementary Table 1) and absence of signiﬁcant change and protein expression in Fxr/ (Figure 6A), but not in BASIC AND TRANSLATIONAL LIVER

/ / Figure 6. 1,25(OH)2D3 treatment lowers plasma and liver cholesterol in Western-dietfed Fxr but not Shp mice. 1,25(OH)2D3-treatment (A) increased Cyp7a1 mRNA and protein expression, (B) decreased plasma and liver cholesterol concentrations, and (C) attenuated hepatic Shp, ileal Fgf15, and increased ileal Asbt mRNA levels in Western-dietfed Fxr/ mice; and Cyp7a1 expression and cholesterol levels [(D)to(E)] were unchanged in 1,25(OH)2D3-treated Western-dietfed Shp/ mice, although ileal Fgf15 mRNA expression was decreased (n ¼ 48) (F); P < . 05: †Western diet vs normal diet; *vehicle-treated, Western-dietfed control vs Western-dietfed, 1,25(OH)2D3-treated mice (Mann-Whitney U test). 1056 Chow et al Gastroenterology Vol. 146, No. 4 RNLTOA LIVER TRANSLATIONAL AI AND BASIC April 2014 VDR Up-Regulates Cyp7a1 and Lowers Cholesterol 1057

Shp/ (Figure 6D) mice, and increased fecal bile acid predicted VDREs are contributing to the repression by excretion of tauro-b-muricholic acid and taurocholate in 1,25(OH)2D3. Deletion analysis confirmed involvement of / / Western-dietfed wild-type and Fxr , but not Shp the 283 and 169 VDREs in the 1,25(OH)2D3-mediated mice, where a higher bile acid pool size but decreased fecal suppression of the SHP promoter. When the major putative excretion composed mostly of lithocholic acid and deoxy- VDREs were tested for their ability to compete with the cholic acid were observed (Supplementary Figure 1C). known interaction of Vdr/Rxra on the rat osteocalcin gene Accordingly, plasma and liver cholesterol were reduced in (rOC-VDRE), excess unlabeled SHP-VDRE (283) abolished / Western-dietfed Fxr mice with 1,25(OH)2D3 treatment binding of the protein complex, and the SHP-VDRE(169) (Figure 6B), but not for plasma (P ¼ .057) and liver only partially competed for Vdr/Rxra binding to the rOC- (P ¼ .23) cholesterol of Western-dietfed Shp/ mice VDRE (Figure 7D); a third weak putative VDRE (250) (Figure 6E). Hepatic Shp mRNA expression was reduced and failed to compete with rOC-VDRE binding. These data are ileal Asbt mRNA returned to basal levels in 1,25(OH)2D3- consistent with the promoter truncation analyses that treated, Western-dietfed Fxr/ mice (Figure 6C), and ileal indicate multiple binding sites are important for the Fgf15 mRNA was decreased in both Western-dietfed, 1,25(OH)2D3-mediated repression of the SHP promoter. / / 1,25(OH)2D3-treated Fxr and Shp mice (Figures 6C and F). EMSA EMSA experiments were conducted with biotinylated Vdr Represses Shp/SHP Promoter Activity SHP (283) and (169) oligonucleotides to test whether To examine if Vdr suppressed Shp/SHP directly, we the Vdr-Rxra complex was binding directly to these sites. performed luciferase reporter assays with proximal Shp/ A distinct protein-DNA complex was formed with SHP SHP promoters of mouse (2 kb) and human (0.5 kb). (283)-VDRE, consistent with direct binding of Vdr/Rxra to CDCA significantly increased Shp/SHP promoter activity in this site (Figure 7E) that was diminished upon addition of the presence of FXR (Figures 7A and B); in addition, unlabeled competitor. No binding was observed for the puta- basal Shp/SHP promoter activation was increased with tive site at 169 (data not shown). Taken together, we pro- 2 co-transfection of the competence factor, Lrh-1. Addition of pose that the 1,25(OH)2D3-mediated suppression of SHP 1,25(OH)2D3 and Vdr strongly repressed Shp/SHP promoter expression occurs through the direct binding of VDR to at least activity, and addition of CDCA and 1,25(OH)2D3 led to Vdr- one DR3 response element in the proximal SHP promoter. mediated repression of Shp/SHP promoter activities that dominated over FXR-mediated activation, and the observation was independent of Lrh-1 (Figure 7A and B). ChIP Assay To identify which residues in human SHP promoter To determine whether VDR binds to the mouse Shp conferred repression by Vdr, we generated a number of promoter in vivo, we performed ChIP with mouse liver sam- truncation mutants in the luciferase (luc) reporter and ples at 12 hours after the fourth dose of 1,25(OH)2D3 treat- tested for loss of repression after addition of 1,25(OH)2D3 ment, whereby Cyp7a1 protein was increased (70%; data not and Vdr (Figure 7C). Sequence analysis revealed 2 putative shown). A significant increase in the recruitment of Vdr and DR3 VDR response elements (VDREs) located within the Rxra to the Shp promoter was found with 1,25(OH)2D3, proximal SHP promoter (at positions 283 and 169; without changes in Lrh-1 recruitment (Figure 7F). We BASIC AND

Supplementary Table 2). The ability of 1,25(OH)2D3 to observed a 1,25(OH)2D3-dependent increase in histone 3 TRANSLATIONAL LIVER repress the SHP 258-luc reporter was diminished lysine 9 trimethylation (H3K9me3, a marker of chromatin compared with that of 311-luc reporter, and abolished in condensation), further supporting an important role for Vdr the 138-luc reporter, consistent with the idea that these 2 in mediating Shp repression (Figure 7F).

= Figure 7. Interaction between Vdr and the SHP promoter. Human embryonic kidney 293 (HEK293) cells were transiently transfected with either (A) mouse Shp (2 kb)-luciferase reporter or (B) human SHP (569 bp)-luciferase reporter in presence or absence of mouse Lrh-1, Vdr, and human FXR and mouse retinoid X receptor a (Rxra). After 6 to 8 hours, cells were treated with vehicle (0.1% EtOH), 50 mM CDCA, 0.5 nM 1,25(OH)2D3,or50mM CDCA þ 0.5 nM 1,25(OH)2D3. Data are mean SD (n ¼ 3). *P < .05, 1,25(OH)2D3 vs vehicle (EtOH) control; CDCAþ1,25(OH)2D3 vs CDCA; CDCAþ1,25(OH)2D3 vs 1,25(OH)2D3.(C)In truncation studies of the SHP promoter, the boxes represent potential VDRE sites, and denotes the construct with putative VDRE (sequence shown) deleted. HEK293 cells were transiently transfected with the indicated SHP promoter luciferase constructs in presence of Lrh-1, Rxra, and Vdr, then treated with vehicle or 0.5 nM 1,25(OH)2D3. Data are mean SD (n ¼ 3); *P < .05 with or without 1,25(OH)2D3 treatment. (D) Vdr/Rxra heterodimers were incubated with 40 nM rOC-VDRE biotin-labeled probe. Where indicated, an unlabeled oligonucleotide competitor (500-fold that of the probe) was added to the reaction mixtures, except for lane 3, where the competitor concentration was a 100-fold that of the probe. (E) Vdr and Rxra nuclear extracts (alone or in combination) were incubated with 40 nM biotin-labeled rOC-VDRE or SHP(-283)-VDRE. Where indicated, the matching unlabeled oligonucleotide competitor (1000-fold of the probe) was added. (F) ChIP assay of Vdr, Rxra, Lrh-1, and H3K9me3 (IgG as control) at the mouse Shp promoter in liver tissues from vehicle- and 1,25(OH)2D3-treated wildtype mice (n ¼ 34/treatment). Quantitative polymerase chain reaction (qPCR) was used to quantify the relative abundance of each species at the Shp promoter. A representative dataset of treated and control mice was shown, and the error bar represents SEM of qPCR triplicates. RLU, relative luciferase unit. 1058 Chow et al Gastroenterology Vol. 146, No. 4

Discussion communication between resident liver cells is also occurring We observed cholesterol lowering in response to in vivo because VDR is also highly expressed and functional in stellate cells.31 We acknowledge that additional 1,25(OH)2D3 treatment, an effect associated with elevated Cyp7a1 mRNA and protein expression and microsomal 1,25(OH)2D3-liganded mechanisms can contribute to activity, with correspondingly larger bile acid pool sizes lowering plasma and liver cholesterol in vivo. Indeed, and/or greater fecal bile acid excretion in hypercholes- 1,25(OH)2D3 treatment of HL-60 macrophages can also terolemic wild-type and Fxr/ mice. We showed that reduce cholesterol by inhibiting human 3-hydroxy-3-methyl- increased Cyp7a1 expression and activity was achieved via glutaryl-CoA reductase activity and increasing acetyl- Vdr-repression of Shp after steady-state treatment of coenzyme A acetyltransferase (ACAT) activity leading to cholesteryl ester accumulation.32 1,25(OH)2D3. The inhibition of Shp and induction of Cyp7a1 by Vdr in mouse livers was clearly demonstrated both The novel mechanism of up-regulation of CYP7A1 after in vivo and in vitro, and our molecular studies showed that 1,25(OH)2D3 treatment suggests that the VDR is a new Vdr activation resulted in repression of Shp/SHP promoter therapeutic target for cholesterol lowering. However, the activities (Figure 7). Through extensive gene profiling of potential utility of this mechanism to treat hypercholester- the ileum and liver of hypercholesterolemic mouse models, olemia is limited due to the dose-limiting hypercalcemia of 25 a 13 we further ruled out the involvement of other transporters, 1,25(OH)2D3 or its precursor, 1 -hydroxyvitamin D3. enzymes, or nuclear receptors known to modulate choles- Use of dietary vitamin D for cholesterol lowering in terol or bile acid processing (Supplementary Figures 2 humans remains somewhat uncertain because only very low and 3). All data point to the Fxr-independent and Shp- levels of 1,25(OH)2D3 are synthesized after ingestion. By contrast, it is not unlikely that vitamin D deficiency would dependent mechanism by which Vdr down-regulates Shp 11 to increase Cyp7a1 in cholesterol lowering. affect cholesterol status. The interplay between the VDR These findings contrast other reports on the down- and cholesterol homeostasis in humans requires continued regulation of mouse hepatic Cyp7a1 mRNA due to Fgf15 investigation with nonhypercalcemic VDR ligands. 10 induction after a high dose of 1,25(OH)2D3. Such divergent results could be explained by differences in the dosing regimen. The notion that VDR is inhibitory to CYP7A1 in Supplementary Material human hepatocytes was based on the absence of substan- Note: To access the supplementary material accompa- tiating evidence on CYP7A1 protein/activity or cholesterol nying this article, visit the online version of Gastroenterology measurements, or timed-matched control samples9; the at www.gastrojournal.org, and at http://dx.doi.org/10. observations were explained as genomic effects arising from 1053/j.gastro.2013.12.027. the interaction between VDR and hepatocyte nuclear factor 4a on the CYP7A1 promoter, or as nongenomic effects via References RNLTOA LIVER TRANSLATIONAL activation of the extracellular signal-regulated kinase 9,28 1. Chiang JY. Regulation of bile acid synthesis: pathways,

AI AND BASIC pathway, although the latter was not reproduced in a 29 nuclear receptors, and mechanisms. J Hepatol 2004; different cell line. It can be argued that conclusions based 40:539–551. solely on in vitro data are debatable because time- 2. Lu TT, Makishima M, Repa JJ, et al. Molecular basis for dependent gene stability exists, and acute cell-based feedback regulation of bile acid synthesis by nuclear studies are expected to be sensitive to differences in treat- receptors. Mol Cell 2000;6:507–515. ment times and conditions. Indeed, in our in vitro human 3. Zöllner G, Marschall HU, Wagner M, et al. Role of nuclear hepatocyte studies, we show time-dependent changes in receptors in the adaptive response to bile acids and CYP7A1 expression (Figure 4B). The involvement of Shp/ cholestasis: pathogenetic and therapeutic consider- SHP may have been easily missed because SHP mRNA has a ations. Mol Pharm 2006;3:231–251. < short half-life ( 30 minutes) due to rapid proteasomal 4. Inagaki T, Choi M, Moschetta A, et al. Fibroblast growth degradation that is under the control of the extracellular factor 15 functions as an enterohepatic signal to regulate 30 signal-regulated kinase pathway. We suggest that longer- bile acid homeostasis. Cell Metab 2005;2:217–225. term effects of steady-state doses of 1,25(OH)2D3 5. Makishima M, Lu TT, Xie W, et al. Vitamin D receptor as an observed in our in vivo studies likely represent physiologic intestinal bile acid sensor. Science 2002;296:1313–1316. responses. 6. Wagner M, Zöllner G, Trauner M. Nuclear receptor In our molecular studies, we conﬁrmed an interaction regulation of the adaptive response of bile acid trans- between the Shp promoter and Vdr protein in the ChIP assay. porters in cholestasis. Semin Liver Dis 2010;30:160–177. The addition of VDR ligand resulted in reduced activity of 7. Honjo Y, Sasaki S, Kobayashi Y, et al. 1,25-Dihydroxyvitamin Shp/SHP promoters (Figure 7), explaining the observed D3 and its receptor inhibit the chenodeoxycholic acid- cholesterol lowering in mouse in vivo. Chromatin remodeling dependent transactivation by farnesoid X receptor. at the Shp promoter in response to 1,25(OH)2D3 is consistent J Endocrinol 2006;188:635–643. with ligand-mediated repression, although the relevant co- 8. Jiang W, Miyamoto T, Kakizawa T, et al. Inhibition of repressor proteins involved in this process have yet to be LXRa signaling by vitamin D receptor: possible role of identiﬁed. Although our data are consistent with a direct role VDR in bile acid synthesis. Biochem Biophys Res Com- for VDR in hepatocytes, we cannot rule out that intercellular mun 2006;351:176–184. April 2014 VDR Up-Regulates Cyp7a1 and Lowers Cholesterol 1059

9. Han S, Chiang JY. Mechanism of vitamin D receptor in- 24. Chow EC, Durk MR, Cummins CL, et al. 1a,25- hibition of cholesterol 7a-hydroxylase gene transcription Dihydroxyvitamin D3 up-regulates P-glycoprotein via in human hepatocytes. Drug Metab Dispos 2009;37: the vitamin D receptor and not farnesoid X receptor in 469–478. both fxr(-/-) and fxr(þ/þ) mice and increased renal and 10. Schmidt DR, Holmstrom SR, Fon Tacer K, et al. Regu- brain efﬂux of digoxin in mice in vivo. J Pharmacol Exp lation of bile acid synthesis by fat-soluble vitamins A and Ther 2011;337:846–859. D. J Biol Chem 2010;285:14486–14494. 25. Chow EC, Quach HP, Vieth R, et al. Temporal changes in 11. Wang JH, Keisala T, Solakivi T, et al. Serum cholesterol tissue 1a,25-dihydroxyvitamin D3, vitamin D receptor target and expression of ApoAI, LXRb and SREBP2 in vitamin D genes, and calcium and PTH Levels after 1,25(OH)2D3 receptor knock-out mice. J Steroid Biochem Mol Biol treatment in mice. Am J Physiol Endocrinol Metab 2013; 2009;113:222–226. 304:E977–E989. 12. Nishida S, Ozeki J, Makishima M. Modulation of bile acid 26. Patel R, Patel M, Tsai R, et al. LXRb is required for metabolism by 1a-hydroxyvitamin D3 administration in glucocorticoid-induced hyperglycemia and hep- mice. Drug Metab Dispos 2009;37:2037–2044. atosteatosis in mice. J Clin Invest 2011;121:431–441.

13. Ogura M, Nishida S, Ishizawa M, et al. Vitamin D3 mod- 27. Zierold C, Mings JA, DeLuca HF. Regulation of 25- ulates the expression of bile acid regulatory genes and hydroxyvitamin D3-24-hydroxylase mRNA by 1,25-dihy- represses inflammation in bile duct-ligated mice. droxyvitamin D3 and parathyroid hormone. J Cell Biochem J Pharmacol Exp Ther 2009;328:564–570. 2003;88:234–237. 14. Wang XX, Jiang T, Shen Y, et al. Vitamin D receptor 28. Han S, Li T, Ellis E, et al. A novel bile acid-activated agonist doxercalciferol modulates dietary fat-induced vitamin D receptor signaling in human hepatocytes. Mol renal disease and renal lipid metabolism. Am J Physiol Endocrinol 2010;24:1151–1164. Renal Physiol 2011;300:F801–F810. 29. Wu FS, Zheng SS, Wu LJ, et al. Calcitriol inhibits the 15. Chow EC, Maeng HJ, Liu S, et al. 1a,25- growth of MHCC97 heptocellular cell lines by down- Dihydroxyvitamin D3 triggered vitamin D receptor and modulating c-met and ERK expressions. Liver Int 2007; farnesoid X receptor-like effects in rat intestine and liver 27:700–707. in vivo. Biopharm Drug Dispos 2009;30:457–475. 30. Miao J, Xiao Z, Kanamaluru D, et al. Bile acid signaling 16. Gascon-Barré M, Demers C, Mirshahi A, et al. The pathways increase stability of small heterodimer partner normal liver harbors the vitamin D nuclear receptor in (SHP) by inhibiting ubiquitin-proteasomal degradation. nonparenchymal and biliary epithelial cells. Hepatology Genes Dev 2009;23:986–996. 2003;37:1034–1042. 31. Ding N, Yu RT, Subramaniam N, et al. A vitamin D re- 17. Ponda MP, Dowd K, Finkielstein D, et al. The short-term ceptor/SMAD genomic circuit gates hepatic fibrotic effects of vitamin D repletion on cholesterol: a random- response. Cell 2013;153:601–613. ized, placebo-controlled trial. Arterioscler Thromb Vasc 32. Jouni ZE, Winzerling JJ, McNamara DJ. 1,25- Biol 2012;32:2510–2515. Dihydroxyvitamin D3-induced HL-60 macrophages: 18. Zittermann A, Frisch S, Berthold HK, et al. Vitamin D regulation of cholesterol and LDL metabolism. Athero- supplementation enhances the beneficial effects of sclerosis 1995;117:125–138. weight loss on cardiovascular disease risk markers. Am J –

Clin Nutr 2009;89:1321 1327. BASIC AND Author names in bold designate shared co-first authorship. 19. Salehpour A, Shidfar F, Hosseinpanah F, et al. Vitamin D3 TRANSLATIONAL LIVER and the risk of CVD in overweight and obese women: a Received December 4, 2012. Accepted December 17, 2013. randomised controlled trial. Br J Nutr 2012:1–8. Reprint requests 20. Rahimi-Ardabili H, Pourghassem Gargari B, Farzadi L. Address requests for reprints to: K. Sandy Pang, PhD, Leslie Dan Faculty of Effects of vitamin D on cardiovascular disease risk fac- Pharmacy, University of Toronto, 144 College Street, Toronto, Ontario, tors in polycystic ovary syndrome women with vitamin D Canada M5S 3M2. e-mail: [email protected]; fax: 416-978-8511. fi – de ciency. J Endocrinol Invest 2013;26:28 32. Acknowledgments 21. Karhapää P, Pihlajamäki J, Pörsti I, et al. Diverse as- The authors thank Drs Nan Wu and Martin Wagner, Baylor College of Medicine, Texas Medical Center, Houston, for assistance with studies on the bile acid sociations of 25-hydroxyvitamin D and 1,25-dihydroxy- pool sizes of Shp/ mice. vitamin D with dyslipidaemias. J Intern Med 2010;268: Han-Joo Maeng’s current affiliation is College of Pharmacy, Inje University, 604–610. 607 Obang-dong, Gimhae, Gyeongnam 621-749, South Korea. Sayeepriyadarshini Anakk’s current affiliation is Department of Molecular 22. Schwartz JB. Effects of vitamin D supplementation in and Integrative Physiology, University of Illinois, Urbana-Champaign, atorvastatin-treated patients: a new drug interaction with Urbana, IL.

an unexpected consequence. Clin Pharmacol Ther 2009; Conﬂicts of interest 85:198–203. The authors disclose no conﬂicts. 23. Davis W, Rockway S, Kwasny M. Effect of a combined Funding therapeutic approach of intensive lipid management, The authors (KSP, CC, ECYC, LM, HPQ, and MRD) gratefully acknowledge omega-3 fatty acid supplementation, and increased support from the Canadian Institutes of Health Research, the Natural Sciences and Engineering Research Council of Canada, and the Ontario serum 25 (OH) vitamin D on coronary calcium scores in Graduate Scholarship Program. SA and DDM were supported by CPRIT asymptomatic adults. Am J Ther 2009;16:326–332. grant RP120138, and R. P. Doherty, Jr. – Welch Chair in Science Q-0022. 1059.e1 Chow et al Gastroenterology Vol. 146, No. 4

Supplementary Methods reduced nicotinamide adenine dinucleotide phosphate (NADPH) generating system to determine the Cyp7a1 ac- Materials tivity according to the high-performance liquid chroma- 1 The 1,25(OH)2D3 was purchased from Sigma-Aldrich tography assay. Canada(Mississauga,ON).Antibodies against Cyp7a1/ Bile acid pool size in hypercholesterolemic CYP7A1 (N-17) were from Santa Cruz Biotechnology mice. Wild-type, Fxr-/- and Shp-/- mice fed the Western (Santa Cruz, CA); those for Gapdh/GAPDH (6C5), Lamin diet for 3 weeks were treated with either vehicle or B1, and VDR (9A7) were from Abcam (Cambridge, MA). 1,25(OH)2D3 (2.5 mg/kg, every other day for 8 days) at the For the ChIP studies, the following antibodies were used: beginning of the third week (each mouse was caged indi- 10 mg of VDR (sc-13133x, Santa-Cruz Biotechnology), vidually). The extraction procedure for bile acid pool was 10 mg of RXR (sc- 774x, Santa-Cruz Biotechnology), 10 mg similar to those described by others.3,4 On the last day of of LRH-1 (H2325, R&D Systems) or 2 mgofH3K9me3 the study, mice were fasted for 4 hours before induction of (ab8898, Abcam) antibodies overnight and used for anesthesia. The intact gall bladder, liver, and intestine immunoprecipitation. Male C57BL/6 mice were obtained were removed altogether. The tissues were reduced to fine from Charles River (Senneville, Quebec, Canada) and Fxr / pieces and placed in a beaker containing 50 mL anhydrous mice (C57BL/6 background) were kind gifts from Dr Frank ethanol. After the addition of 50 mL of the internal stan- J. Gonzalez (National Institutes of Health, Bethesda, MD). dard (1 mg/mL chenodeoxycholic acid-D4 [CDCA-d4]in The Shp / mice (C57BL/6 background), aged 8 to 12 methanol, C/D/N Isotopes; Pointe-Claire, Quebec, Canada), weeks, were from the laboratory of Dr David M. Moore the content was boiled at 80C for 1 hour. After cooling, (Baylor College of Medicine, Houston, TX). Studies were the extracts were filtered through a Whatman filter paper performed in accordance with institutionally approved and brought up to 50 mL in a volumetric flask with animal protocols. anhydrous ethanol. Then 500 mL of the extract was centrifuged and filtered through an Ultra-free-MC centrif- Plasmids ugal filter device containing 0.22-mm polyvinylidene fl The pCMX, pCMX-hRXRa, pCMX-mRXRa, pCMX-mLRH-1, di uoride membrane (Millipore, Billerica, MA) before pGEM, pCMX-b-galactosidase, hSHP(569)-luc, and hSHP analysis. Standard solutions containing known amounts (371)-luc were from Dr David Manglesdorf (University of of bile acids were processed and extracted in the Texas Southwestern Medical Center, Dallas, TX), pEF-mVDR same manner. Samples were then analyzed by liquid was from Dr Rommel G. Tirona (University of Western chromatography-tandem mass spectrometry using 6410 Ontario, London, ON), and mSHP promoter was from Dr Li Triple Quad LC/MS instrument (Agilent Technologies) with Wang (University of Utah, UT). Human SHP promoter Electrospray ionization (ESI) source in negative ion mode fi 5 deletion constructs were generated by polymerase chain as described previously, with slight modi cations. Sam- m reaction (PCR) amplification. The PCR fragments were ples (1 L) were separated on a Zorbax XDB-C18 column m ligated into the HindIII and BglII sites of the luciferase (4.6 50 mm, 3.5 m) with a C18 guard column at 0.4 reporter pGL3 (Promega, Madison, WI) to generate mL/min. The mobile phase consisted of high-performance hSHP(238)-luc and hSHP(138)-luc. liquid chromatography grade water/10 mM NH4Ac/ 0.024% formic acid (Solvent A) and methanol/0.024% formic acid (Solvent B). A gradient was utilized over 30 Real-Time PCR minutes: 015 minutes, 70%80% (Solvent B); 1517 Primer sequences are described in Supplementary minutes, 80% (Solvent B); 1720 minutes, 80%95% Methods Table 1. Total mRNA extraction and quantitative 1 (Solvent B); 20 26 minutes, 95% (Solvent B). Mass spec- PCR procedures are previously described. mRNA levels are trometry parameters were as follows: gas temperature normalized to cyclophilin (for mouse liver and hepatocytes), 350C, nebulizer pressure 35 psi, drying gas (nitrogen) GAPDH (for human hepatocytes), or villin (mouse intestine) 12 L/min, VCap 6000 V (negative), and column tempera- then expressed as relative mRNA expression of the control. ture 40C. The fragmentor voltage was 200 V and collision energy was 5 V for all the compounds monitored. Selective Western Blotting ion monitoring was used to detect the conjugated and Vdr and Cyp7a1/CYP7A1 protein expression was unconjugated bile acids (Supplementary Methods Table 2). determined by Western blotting methods, as described,1,2 Bile acids were quantified based on peak areas using after loading of 50 mg total protein samples onto 10% so- external calibration curves of standards prepared in dium dodecyl sulfate-polyacrylamide gels and transferring methanol. CDCA-d4 was used to calculate the recovery of to nitrocellulose membranes.1 bile acids after extraction relative to a blank control. Liver nuclear protein, microsomes, and Plasma and tissue cholesterol. Total plasma Cyp7a1 activity. Liver nuclear protein and microsomes cholesterol was determined by the Total Cholesterol Kit for Western blotting were described.1 For activity assays, (Wako Diagnostics, Richmond, VA). For liver cholesterol mouse liver microsomes (2 mg) obtained from sequential measurements, lipids were extracted from approximately centrifugation were incubated with cholesterol and a 0.2 g liver, homogenized in chloroform:methanol (2:1, v/v), April 2014 VDR Up-Regulates Cyp7a1 and Lowers Cholesterol 1059.e2

6 as described previously, and cholesterol concentrations Supplementary References were determined from extracts using Infinity Cholesterol a reagents (Thermo Scientific, Rockford, IL). 1. Chow EC, Maeng HJ, Liu S, et al. 1 ,25-Dihydroxyvitamin Determination of tissue 1a,25-dihydroxy- D3 triggered vitamin D receptor and farnesoid X receptor- like effects in rat intestine and liver in vivo. Biopharm Drug vitamin D3. The 1,25(OH) D concentrations in mouse 2 3 Dispos 2009;30:457–475. plasma and liver were assayed using an enzyme- 2. Chow EC, Durk MR, Cummins CL, et al. 1a,25- immunoassay kit.7 Briefly, weighed liver samples were Dihydroxyvitamin D up-regulates P-glycoprotein via the added to double-distilled water up to 1 mL. The sample was 3 vitamin D receptor and not farnesoid X receptor in both homogenized with 3.75 mL of a mixture of methylene chlo- fxr(-/-) and fxr(þ/þ) mice and increased renal and brain ride and methanol (1:2 v/v). Then 1.25 mL of methylene efflux of digoxin in mice in vivo. J Pharmacol Exp Ther chloride was added, mixed for 1 minute, followed by addition 2011;337:846–859. of 1.25 mL double-distilled water and mixed for another 3. Turley SD, Schwarz M, Spady DK, et al. Gender-related minute, before centrifugation at 3000 rpm for 20 minutes at differences in bile acid and sterol metabolism in outbred room temperature. The extractant (bottom phase) was CD-1 mice fed low- and high-cholesterol diets. Hepatol- retrieved by a glass syringe metal needle set. The extraction ogy 1998;28:1088–1094. procedure was repeated with another 1.25 mL methylene 4. Zhang Y, Breevoort SR, Angdisen J, et al. Liver LXRa chloride. The recovered, bottom extractant was pooled with expression is crucial for whole body cholesterol homeo- that from the previous extraction, dried under N2, and stasis and reverse cholesterol transport in mice. J Clin reconstituted in 0.3 mL charcoal-stripped human serum and Invest 2012;122:1688–1699. analyzed by the enzyme immunoassay kit (Immunodiag- 5. Lee YK, Schmidt DR, Cummins CL, et al. Liver receptor nostics Systems Inc., Scottsdale, AZ). homolog-1 regulates bile acid homeostasis but is not Human hepatocytes. Fresh, human primary hepa- essential for feedback regulation of bile acid synthesis. tocytes from 3 donors (donor ID# Hu1177, Hu1210, Mol Endocrinol 2008;22:1345–1356. Hu1284) were supplied by Dr Jasminder Sahi (Life Tech- 6. Patel R, Patel M, Tsai R, et al. LXRb is required for nologies; parent company of Invitrogen) as kind gifts. Hu- glucocorticoid-induced hyperglycemia and hep- man hepatocytes (donor ID# Hu1284, male, Caucasian, age atosteatosis in mice. J Clin Invest 2011;121:431–441. 51 years), which showed a stable and high VDR and CYP7A1 7. Chow EC, Quach HP, Vieth R, et al. Temporal changes in mRNA expression (CT value approximately 26 27 and tissue 1a,25-dihydroxyvitamin D3, vitamin D receptor approximately 2324, respectively), were treated with target genes, and calcium and PTH levels after 1, 1,25(OH)2D3 and were harvested at various time points to 25(OH)2D3 treatment in mice. Am J Physiol Endocrinol examine changes in mRNA and protein expression. Metab 2013;304:E977–E989. 1059.e3 Chow et al Gastroenterology Vol. 146, No. 4

Supplementary Table 1.Plasma Calcium, Phosphorus, ALT, and Portal Bile Acid Concentrations in Wild-Type (Fxrþ/þ) and Fxr/ Mice Fed a Normal Diet (ND) and in Wild-Type, Fxr/, and Shp/ Mice Fed a Western Diet (WD)

Portal bile acid Plasma calcium, mg/dL Plasma phosphorus, mg/dL Plasma ALT, IU/mL concentration, mM

Wild-type Vehicle controlND 9.6 0.2 19.2 0.9 10.2 0.9 28.5 7.1 a 1,25(OH)2D3 treatedND 12.5 0.2 18.7 0.7 11.4 1.2 32.4 9.5 Fxr/ Vehicle controlND 8.1 0.4b 19.8 1.3 144 19b 53.4 6.8 a a 1,25(OH)2D3 treatedND 10.0 0.6 17.3 0.8 32.4 4.7 65.0 11.1 Wild-type Vehicle controlND 9.1 0.1 17.3 0.9 7.2 1.6 33.8 14.9 Vehicle controlWD 9.0 0.1 21.1 1.0c 23.5 2.3c 31.8 6.6 d 1,25(OH)2D3 treatedWD 12.8 0.5 19.5 0.6 17.2 2.7 31.0 2.6 Fxr/ Vehicle controlND 8.1 0.4 19.8 1.3 96.3 9.0 27.6 6.5 Vehicle controlWD 11.4 0.44c 20.9 0.75 81.4 59.9 62.7 10.0c d 1,25(OH)2D3 treatedWD 17.1 0.59 21.6 0.86 89.2 54.3 80.1 25.5 Shp/ Vehicle controlND 9.3 0.2 15.8 1.1 12.0 2.0 28.3 9.0 Vehicle controlWD 11.0 0.6c 24.3 2.2c 23.7 4.5 56.0 10.7 d d 1,25(OH)2D3 treatedWD 13.7 0.4 16.2 1.4 46.2 14.7 101 37.3

NOTE. Plasma was diluted 350-fold with 1% HNO3 and calcium and phosphorus were determined by inductively coupled plasma atomic emission spectroscopy. Plasma ALT and serum portal bile acid concentrations were determined by ALT kit (Bioquant, Nashville, TN) and the total bile acids assay kit (Diazyme, Poway, CA), respectively. Mice fed a normal diet (ND) showed approximately 23% to 30% increase in calcium when treated with 1,25(OH)2D3. The same was observed in Western Diet (WD)-fed Fxr/ and Shp/ mice vs their untreated WD-fed counterparts. There was no dramatic change in plasma phosphorous and ALT levels, and total serum bile acid (portal) concentration with 1,25(OH)2D3 treatment. Data represented mean SEM (n ¼ 48). ALT, alanine aminotransferase. aP < .05, compared with vehicle control (Mann-Whitney U test). bP < .05, compared with Fxrþ/þ control (Mann-Whitney U test). cP < .05, compared with normal diet vehicle control (Mann-Whitney U test). dP < .05, compared with Western diet vehicle control (Mann-Whitney U test).

Supplementary Table 2.Oligonucleotide Sequences for EMSA

Oligonucleotide 50/ 30

rOC-VDRE(þ) GCACTGGGTGAATGAGGACATTAC hSHP(283)-VDRE(þ) GTTAATGACCTTGTTTATCCACTTG hSHP(250)-VDRE() GATAAGGGGCAGCTGAGTGAGCGGC hSHP(169)-VDRE(þ) CGTGGGGTTCCCAATGCCCCCTCCC

H, human. April 2014 VDR Up-Regulates Cyp7a1 and Lowers Cholesterol 1059.e4

Supplementary Methods Table 1.Mouse and Human Primer Sequences for Quantitative Polymerase Chain Reaction

Gene bank no. Forward (50/ 30 sequence) Reverse (50/ 30 sequence) mVdr NM_009504 GAGGTGTCTGAAGCCTGGAG ACCTGCTTTCCTGGGTAGGT mFxr NM_009108 CGGAACAGAAACCTTGTTTCG TTGCCACATAAATATTCATTGAGATT mShp NM_011850 CAGCGCTGCCTGGAGTCT AGGATCGTGCCCTTCAGGTA mLrh-1 NM_001159769 CCCTGCTGGACTACACGGTTT CGGGTAGCCGAAGAAGTAGCT mLxra NM_013839 GGATAGGGTTGGAGTCAGCA GGAGCGCCTGTTACACTGTT mHnf-4a NM_008261 CCAAGAGGTCCATGGTGTTTAAG GTGCCGAGGGACGATGTAGT mFgf15 NM_008003 ACGGGCTGATTCGCTACTC TGTAGCCTAAACAGTCCATTTCCT mIbabp NM_008375.1 CAAGGCTACCGTGAAGATGGA CCCACGACCTCCGAAGTCT mApoE NM_009696.3 AAGCAACCAACCCTGGGAG TGCACCCAGCGCAGGTA mVldlr NM_001161420.1 GAGCCCCTGAAGGAATGCC CCTATAACTAGGTCTTTGCAGATATGG mLdlr NM_010700.2 AGGCTGTGGGCTCCATAGG TGCGGTCCAGGGTCATCT mSr-b1 NM_016741.1 GGGAGCGTGGACCCTATGT CGTTGTCATTGAAGGTGATGT mCyp7a1 NM_007824 AGCAACTAAACAACCTGCCAGTACTA GTCCGGATATTCAAGGATGCA mCyp8b1 NM_010012.3 GCCTTCAAGTATGATCGGTTCCT GATCTTCTTGCCCGACTTGTAGA mCyp24a1 NM_009996 CTGCCCCATTGACAAAAGGC CTCACCGTCGGTCATCAGC mCyp27a1 NM_024264.4 CTGCGTCAGGCTTTGAAACA TCGTTTAAGGCATCCGTGTAGA mHMG CoA Reductase NM_008255.2 CAAGGAGCATGCAAAGACAA GCCATCACAGTGCCACATAC mNpc1l1 NM_207242.2 TGGACTGGAAGGACCATTTCC GCGCCCCGTAGTCAGCTAT mAbca1 NM_013454.3 CGTTTCCGGGAAGTGTCCTA CTAGAGATGACAAGGAGGATGGA mAbcg5 NM_031884.1 TCAATGAGTTTTACGGCCTGAA GCACATCGGGTGATTTAGCA mAbcg8 NM_026180.2 TGCCCACCTTCCACATGTC ATGAAGCCGGCAGTAAGGTAGA mBsep NM_021022.3 ACAGCACTACAGCTCATTCAGAG TCCATGCTCAAAGCCAATGATCA mAsbt NM_011388 GATAGATGGCGACATGGACCTC CAATCGTTCCCGAGTCAACC mNtcp NM_011387.2 ATCTGACCAGCATTGAGGCTC CCGTCGTAGATTCCTTGCTGT mOsta NM_145932.3 TACAAGAACACCCTTTGCCC CGAGGAATCCAGAGACCAAA mOstb NM_178933.2 GTATTTTCGTGCAGAAGATGCG TTTCTGTTTGCCAGGATGCTC mVillin NM_009509 TCCTGGCTATCCACAAGACC CTCTCGTTGCCTTGAACCTC mCyclophillin X58990 GGAGATGGCACAGGAGGAA GCCCGTAGTGCTTCAGCTT hCYP7A1 NM_000780.3 GAATGCTGGTCAAAAAGTC TGAAATCCTCCTTAGCTGT hCYP24A1 NM_000782 CAGCGAACTGAACAAATGGTCG TCTCTTCTCATACAACACGAGGCAG hGAPDH NM_002046 GAAGGTGAAGGTCGGAGTC GAAGATGGTGATGGGATTTC h, human; m, mouse.

Supplementary Methods Table 2.Summary of Liquid Chromatography-Tandem Mass Spectrometry Parameters Used to Quantify Bile Acids

Retention Ion monitored Compound time, min [M-H], m/z

Taurocholic acid: tCA 10.4 514/514 Cholic acid: CA 21.2 407/407 Deoxycholic acid: DCA/Chenodeoxycholic acid: CDCA 23.7 391/391 Tauro-b-muricholic acid: tb-MCA 4.3 514/514 Lithocholic acid: LCA 24.3 375/375 Chenodeoxycholic acid-d4: CDCA-d4 23.2 395/395 1059.e5 Chow et al Gastroenterology Vol. 146, No. 4 April 2014 VDR Up-Regulates Cyp7a1 and Lowers Cholesterol 1059.e6

Supplementary Figure 1. 1,25(OH)2D3 treatment differentially alters bile acid pool size and fecal bile acid excretion in wildtype, Fxr -/- and Shp-/- mice. The intestinal and liver genes that affect bile acid processing were first compared among normal diet (ND)-fed wildtype and knockout mice; relative values for wildtype mice were set as unity (A) Absence of Fxr in Fxr -/- mice led to reduced Shp, Fgf15, Ibabp, and Ost-a and Ost-b in ileum and lower Shp but increased Abcg8 and Cyp7a1 mRNA expression in liver compared to those of wildtype mice. (B) Absence of Shp in Shp-/- mice led to reduced Fxr in ileum and increased Vldlr, Sr-b1, Abcg5, Abcg8, Cyp7a1, Cyp8b1, and HMG CoA reductase mRNA in liver, as found previously.1,2 *, P < .05, Mann-Whitney U test: between wildtype and Fxr -/-, ND-fed mice (C) Individual bile acids were quantified by LC/MS using external calibration curves of pure bile acid standards and CDCA-d4 as an internal standard. “Others” represents the sum of CDCA, CA, LCA, and DCA. In wildtype mice, the Western diet (WD) did not alter amounts of individual bile acids nor their sum (bile acid pool size), but the total bile acid pool size and fecal excretion of bile acids were increased upon treatment of -/- 1,25(OH)2D3 (when Cyp7a1 was increased). In Fxr mice, 1,25(OH)2D3 treatment did not significantly increase the total bile acid pool size, however, the fecal excretion of total bile acids and of tb-MCA and tCA was disproportionately increased relative -/- to the modest change in pool size. We speculate that Fxr mice, despite showing increased Cyp7a1 with 1,25(OH)2D3 treatment, do not reabsorb bile acids as efficiently as wildtype mice. Their lower basal expression of intestinal Osta-Ostb (30- 40% of wildtype mice) and virtually non-existent Ibabp level (see Supplementary Figures 1A and 3A) may give rise to a net, lower reabsorption/reclamation of bile acids and a disproportionately higher fecal excretion of tb-MCA and tCA. In -/- 1,25(OH)2D3-treated Shp mice, the total bile acid pool size was increased, whereas fecal bile acid excretion was dramatically -/- decreased. The Shp mice are distinct because the fecal bile acids excreted after 1,25(OH)2D3 treatment reflect primarily -/- products of bile acid conversion by bacteria in the colon (LCA and DCA), whereas in wildtype and Fxr mice, 1,25(OH)2D3 treatment significantly increased the levels of the bile acid tb-MCA in the feces. The difference in the composition of the fecal bile acids under different treatment conditions suggest that, despite the increased bile acid pool size in response to -/- 1,25(OH)2D3 in the Shp mouse, this is unlikely to be due to changes in bile acid metabolizing enzymes. P < .05, Mann- Whitney U tests: †, for WD- vs ND-fed mice of same genotype; *, between vehicle vs 1,25(OH)2D3 treated WD-fed, mice of same genotype. 1059.e7 Chow et al Gastroenterology Vol. 146, No. 4

Supplementary Figure 2. Changes in mRNA expression of other cholesterol related genes in (A) ileum and (B) liver of WD-fed wildtype mice treated with 1,25(OH)2D3. A set of vehicle-treated wildtype mice on ND served as controls. (A) Ileal Shp, Abca1, Abcg5, and Abcg8 mRNA expressions were elevated by the WD, and mRNA expression levels of ileal Abca1 and Abcg5 mRNA was decreased upon exposure to 1,25(OH)2D3. (B) Hepatic Sr-b1, Abcg5, and Abcg8 mRNA expressions were increased whereas those for FXR, Vldlr and HMG Co-A reductase were decreased by the WD. Upon 1,25(OH)2D3 treatment, mRNA expression of ApoE, Sr-b1, Abcg5, and Abcg8 were signiﬁcantly decreased, while Vldlr mRNA was increased back to normal. The symbols † and * denote signiﬁcant differences (P < .05) using Mann-Whitney U test between the ND-fed and WD-fed controls, and between the WD-fed 1,25(OH)2D3-treated vs WD-fed, vehicle-treated control, respectively. Data are the mean SEM (n ¼ 4-8).

Supplementary Figure 3. Changes in mRNA expression in the (A) ileum (B) livers of WD-fed Fxr -/- and Shp -/- mice treated with 1,25(OH)2D3; these were compared against the relative abundances for ND-fed wildtype mouse (Supplementary Figure 1) assigned as unity for comparison. In Fxr -/- mice, (A) mRNA expression of ileal Abca1, Abcg5, and Abcg8 was elevated by the WD, but returned to basal levels upon treatment of 1,25(OH)2D3, whereas Ileal Npc1l1 mRNA expression was slightly decreased by the WD. (B) mRNA expression levels of hepatic Vdr, Lxra, Lrh-1, Bsep, ApoE, Abca1, Abcg5, and Abcg8 were all increased by the WD whereas Cyp8b1 and HMG Co-A reductase mRNA levels were decreased; Ntcp was unchanged. Upon -/- 1,25(OH)2D3 treatment, mRNA expression levels of Lxra, ApoE, Abca1, Abcg8 and Bsep were decreased in WD-fed Fxr mice. (C) In Shp -/- mice, ileal mRNA expression of Npc1l1 was slightly decreased and that of Abca1 was elevated by the WD. -/- Upon treatment with 1,25(OH)2D3, both ileal Npc1l1 and Abca1 mRNA level returned to basal levels in WD-fed Shp mice. (D) In Shp -/- mice, hepatic Bsep, Abca1 and Abcg5 mRNA expression was elevated slightly by the WD while Ldlr and HMG Co-A reductase mRNA levels were decreased. Upon 1,25(OH)2D3 treatment, mRNA expression levels of Sr-b1, Abcg5 and Bsep were decreased, but there was no change in HMG CoA reductase, Bsep, or Ntcp. The symbols † and * denote signiﬁcant differences (P < .05) using Mann-Whitney U test between the ND-fed and WD-fed controls, and between the WD-fed 1,25(OH)2D3-treated vs WD-fed, vehicle-treated control, respectively. Data represent the mean SEM (n ¼ 4-8). 1521-009X/44/2/189–208$25.00 http://dx.doi.org/10.1124/dmd.115.067033 DRUG METABOLISM AND DISPOSITION Drug Metab Dispos 44:189–208, February 2016 Copyright ª 2016 by The American Society for Pharmacology and Experimental Therapeutics Physiologically-Based Pharmacokinetic-Pharmacodynamic

Modeling of 1a,25-Dihydroxyvitamin D3 in Mice Vidya Ramakrishnan,1 Qi Joy Yang,1 Holly P. Quach, Yanguang Cao,2 Edwin C. Y. Chow, Donald E. Mager,1 and K. Sandy Pang1

Department of Pharmaceutical Sciences, Leslie Dan Faculty of Pharmacy, University of Toronto, Toronto, Ontario, Canada (Q.J.Y., H.P.Q., E.C.Y.C., K.S.P.); and Department of Pharmaceutical Sciences, University at Buffalo, SUNY, Buffalo, New York (V.R., Y.C., D.E.M.)

Received August 31, 2015; accepted November 18, 2015

ABSTRACT

1a,25-Dihydroxyvitamin D3 [1,25(OH)2D3] concentrations are regu- segregated flow model (SFM) that describes a low and partial lated by renal CYP27B1 for synthesis and CYP24A1 for degradation. intestinal (blood/plasma) flow to enterocytes was nested within both Downloaded from

Published plasma and tissue 1,25(OH)2D3 concentrations and mRNA models for comparison with the traditional model for intestine (TM) fold change expression of Cyp24a1 and Cyp27b1 following repetitive where the entire flow perfuses the intestine. Both the mPBPK(SFM)- i.p. injections to C57BL/6 mice (2.5 mg 3 kg21 every 2 days for 4 PD and full PBPK(SFM)-PD models described the i.p. plasma and doses) were fitted with a minimal and full physiologically-based tissue 1,25(OH)2D3 concentrations and fold changes in mRNA pharmacokinetic-pharmacodynamic models (PBPK-PD). The mini- expression significantly better than the TM counterparts with F test mal physiologically-based pharmacokinetic-pharmacodynamic comparisons. The full PBPK(SFM)-PD fits showed estimates with dmd.aspetjournals.org linked model (mPBPK-PD) related Cyp24a1 mRNA fold changes to good precision (lower percentage of coefficient of variation), and the linear changes in tissue/tissue baseline 1,25(OH)2D3 concentration model was more robust in predicting data from escalating i.v. doses ratios, whereas the full physiologically-based pharmacokinetic- (2, 60, and 120 pmol) and the rebound in 1,25(OH)2D3 tissue pharmacodynamic model (PBPK-PD) related measured tissue concentrations after dosing termination. The full PBPK(SFM)-PD

Cyp24a1 and Cyp27b1 fold changes to tissue 1,25(OH)2D3 concen- model performed the best among the tested models for describing trations with indirect response, sigmoidal maximal stimulatory the complex pharmacokinetic-pharmacodynamic interplay among

effect/maximal inhibitory effect functions. Moreover, the intestinal Cyp27b1, Cyp24a1, and 1,25(OH)2D3. at ASPET Journals on June 27, 2016

Introduction or CYP27B1, the rate-limiting enzyme in kidney to form the active Vitamin D is a collection of fat-soluble prohormone steroids with a ligand, 1a,25-dihydroxyvitamin D3 [1,25(OH)2D3](Jonesetal., diverse range of biologic effects that are primarily endocrine in nature 1998). In circulation, 1,25(OH)2D3 is present at very low concentrations that are not readily monitored by conventional means; the (Norman et al., 1992). The two major lipophilic forms are vitamin D3/ 21 25(OH)D3 concentration (30 ng mL ) is 1000-fold higher compared cholecalciferol and vitamin D2/ergocalciferol, which are extracted from food sources and produced upon sun exposure to the skin and activation with 1,25(OH)2D3 and is used to determine the vitamin D status (Holick, of 7-dehydrocholesterol. The two forms exist bound to the vitamin 2009). The active ligand, 1,25(OH)2D3, binds to the vitamin D receptor D–binding protein in plasma and are activated sequentially via hydroxyl- (VDR) in multiple tissues throughout the body to regulate the expression ation, first in liver to form the circulating metabolite 25-hydroxyvitamin of genes relating to various biologic processes (Jones et al., 1998) and plays a vital role in regulating the calcium-phosphate mineral balance by D3 [25(OH)D3] by CYP2R1 and CYP27A1, and then by 1a-hydroxylase enhancing calcium and phosphate absorption by the intestine. Active 1,25(OH)2D3 is known to exhibit antiproliferative, immunosuppressive, and anti-inflammatory effects (Clemens et al., 1983; Lemire, 2000; This work was supported by the Canadian Institutes of Health Research (to K.S.P.), Topilski et al., 2004). New, potential therapeutic targets of the VDR-bound the National Sciences and Engineering Research Council of Canada (to H.P.Q. and 1,25(OH)2D3, relating to cholesterol (Chow et al., 2014)- and cerebral beta- E.C.Y.C.), and the Ontario Graduate Scholarship Program (to H.P.Q. and Q.J.Y.). – 1V.R. and Q.J.Y. are co-first authors, and D.E.M. and K.S.P. are co-senior authors. amyloid (Durk et al., 2014) lowering properties, have also been reported. 2Division of Pharmacotherapy and Experimental Therapeutics, Eshelman School Processes controlling the disposition of 1,25(OH)2D3 are complex. of Pharmacy, University of North Carolina, Chapel Hill, NC 27599. Endogenous concentrations of 1,25(OH)2D3 in circulation and tissues dx.doi.org/10.1124/dmd.115.067033. are tightly regulated, especially its synthesis by CYP27B1 and

ABBREVIATIONS: 25(OH)D3, 25-hydroxyvitamin D3; 1,25(OH)2D3,1a,25-dihydroxyvitamin D3;Cmax, peak concentration; Emax, maximal stimulatory effect; FC, fold change; fd, fraction of cardiac output (blood or plasma) with Fick’s law of perfusion; fQ, fraction of intestinal flow perfusing the enterocyte region; Hct, hematocrit; Imax, maximal inhibitory effect; KT, tissue to plasma partitioning ratio; MAPE, median absolute prediction error; mPBPK-PD, minimal physiologically-based pharmacokinetic-pharmacodynamic linked model; MPE, median prediction error; PBPK, physiologically-based pharmacokinetic model; PBPK-PD, physiologically-based pharmacokinetic-pharmacodynamic model; PE, prediction error; PK/PD, pharmacokinetic/pharmacodynamic; PO, by-mouth; QCO, plasma cardiac output; QT, plasma flow to tissue; Rsyn, net synthesis rate; SFM, segregated flow model for intestine; TM, traditional model for intestine; VDR, vitamin D receptor.

189 190 Ramakrishnan et al. degradation by CYP24A1. CYP27B1, the rate-limiting synthetic 1,25(OH)2D3 (Quach et al., 2015), suggesting the need to simultaneously enzyme that tightly regulates 1a-hydroxylation of 25(OH)D3 to form incorporate pharmacodynamics into modeling. In this study, we 1,25(OH)2D3, is inhibited by the 1,25(OH)2D3-bound VDR via the revisited the rich data obtained from repeated i.p. injections of calcium-sensing receptor and parathyroid hormone (Shinki et al., 1992; 1,25(OH)2D3 to mice, which included 1,25(OH)2D3 tissue concentrations Lemay et al., 1995). CYP24A1 or 25-hydroxyvitamin D-24-hydroxylase and fold changes (FC; ratio of changed/basal mRNA level) of renal (24-hydroxylase) is a mitochondrial P450 enzyme that catalyzes the Cyp27b1 and Cyp24a1 mRNA expressions in kidney, ileum, liver, and hydroxylation of both 25(OH)D3 and 1,25(OH)2D3 at carbon-24 to form brain (Chow et al., 2013). We examined the utility of a minimal 24,25(OH)2D3 and 1,24,25(OH)3D3 (Jones et al., 1998; Henry, 2001). physiologically based pharmacokinetic model (mPBPK-PD) to parsimo- This inactivation pathway eventually leads to the production of the more niously describe tissue concentrations versus time, a model that was polar metabolite, calcitroic acid, and induces elimination of 1,25(OH)2D3 originally designed to use plasma concentrations and tissue to plasma and (St-Arnaud, 1999). Upregulation of CYP24A1 by 1,25(OH)2D3 exerts blood partition coefficients (KT), flow terms expressed in terms of the feedback control to reduce 1,25(OH)2D3 concentrations and is a hallmark fraction of cardiac output with Fick’s law of perfusion (fd), and intrinsic of 1,25(OH)2D3 upregulation. clearances (Cao and Jusko, 2012); the pharmacodynamic component was Few mathematical models exist to relate these complex interac- simplified by relating Cyp24a1 FCs to 1,25(OH)2D3 tissue/baseline tions on the absorption, distribution, metabolism, and elimination of tissue concentration ratios. We also employed a full physiologically- 1,25(OH)2D3. A recent compartmental model showed that changes in based pharmacokinetic-pharmacodynamic model (PBPK-PD), consist- the pharmacodynamics of Cyp27b1 and Cyp24a1, when incorporated ing of the same number of “lumped” tissue compartments as the mPBPK- into the pharmacokinetic model, greatly improved description of the PD model, to fit to the full set of plasma and tissue 1,25(OH)2D3 Downloaded from kinetic profiles of 1,25(OH)2D3 in mice after increasing i.v. doses of concentration-time data, together with FC of Cyp24a1 and Cyp27b1 dmd.aspetjournals.org at ASPET Journals on June 27, 2016

Fig. 1. Schematic presentation of the mPBPK-PD models for 1,25(OH)2D3 kinetics in mice. QT and VT denote the plasma flow rates, and tissue volume, respectively, representing the plasma and tissue [brain (Br), kidney (K), liver (L), ileum (I), and peripheral (peri) or other tissue] compartments. fQ is the fractional intestinal (blood or plasma) flow perfusing the enterocyte region. CLint,met,T is the metabolic intrinsic clearance in tissue; ka and kdeg are the first-order absorption and degradation rate constants in the intestinal lumen, respectively. kin,Cyp24a1,T and kout,Cyp24a1,T denote the turnover rate constants for Cyp24a1 in subcompartments in various tissues. An assumption made was that the i.p. dose was absorbed solely by the intestine. See Tables 1 and 2 for detailed description of assigned and fixed parameters used in the models. For the SFM, the intestine was viewed as two tissue subcompartments (serosa and enterocytes) perfused by the serosal (70–95% total intestinal flow) and enterocyte (5–30% total intestinal flow) flow (right panel). This intestinal unit may be substituted into the mPBPK(TM)-PD model to obtain the mPBPK(SFM)-PD model. PBPK-PD Modeling of 1,25(OH)2D3 in Mice 191 mRNA expressions, using indirect response models consisting of DB00136) concentrations and FC in mRNA expressions of Cyp24a1 in various maximal stimulatory effect (Emax) [or maximal inhibitory effect (Imax)] tissues and for Cyp27b1 in kidney were obtained from previously published 21 and EC (or IC ) terms. Moreover, the subtleties of intestinal tissue in vivo pharmacokinetic studies. In these studies, 0 (control) or 2.5 mg kg 50 50 21 perfusion patterns that describe route-dependent intestinal metabolism 1,25(OH)2D3 (120 pmol or 0.05 mg mouse ), dissolved in sterile corn oil, and distribution were compared, as follows: the traditional intestine was administered i.p. to male C57BL/6 mice (8 weeks old) every other day over 8 days, or q2d 4 (Chow et al., 2013). model (TM) in which the entire blood flow perfuses the intestine tissue as a whole; the segregated flow model (SFM) that describes the intestine as the The PBPK-PD Models enterocyte region, perfused by a low and partial intestinal blood flow (5– 30%); and a serosal region, perfused by the remaining flow (Cong et al., The mPBPK-PD Model. Cao and Jusko (2012) described a minimal 2000). The SFM describes a greater extent of intestinal elimination with physiologically-based pharmacokinetic model (PBPK) with a minimum number of compartments with extensive lumping: the plasma, the liver, and two lumped oral compared with i.v. dosing and delimits access of drug after i.v. dosing compartments, together with K as the fitted constant. The plasma flows (Q ) are to enterocytes due to the low flow to that region (Cong et al., 2000). T T expressed as fdQCO for the lumped tissue compartments, with fd as the fraction of plasma cardiac output (QCO) to the lumped compartment. Our mPBPK-PD Materials and Methods consists of 11 compartments, with 6 representing various tissues [plasma, brain, liver, kidney, ileum, and peripheral (or other) compartments] that are inter- Published 1,25(OH) D Data and FC in mRNA Expressions 2 3 connected in a physiologically relevant manner. Four subcompartments cor-

Data used for the modeling of plasma and tissue 1,25(OH)2D3 (or calcitriol, responding to Cyp24a1 enzyme in liver, kidney, ileum, and brain were used chemical structure obtainable from DrugBank: http://www.drugbank.ca/drugs/ to account for the synthesis and degradation of the enzyme. For simplicity, Downloaded from dmd.aspetjournals.org at ASPET Journals on June 27, 2016

Fig. 2. Schematic presentation of the full PBPK-PD models for 1,25(OH)2D3 kinetics in mice. QT and VT denote the plasma flow rates and tissue volume, respectively, representing the plasma and tissue [brain (Br), kidney (K), liver (L), ileum (I), and peripheral (peri) or other tissue] compartments. fQ is the fractional intestinal (blood or plasma) flow perfusing the enterocyte region. CLint,met,T is the metabolic intrinsic clearance in tissue; ka and kdeg are the first-order absorption and degradation rate constants in the intestinal lumen, respectively. kin,Cyp24a1,T,kout,Cyp24a1,T,kin,Cyp27a1,K, and kout,Cyp27b1,K denote the turnover rate constants for Cyp24a1 and Cyp27b1, respectively, in subcompartments in various tissues and kidney. An assumption made was that the i.p. dose was absorbed solely by the intestine. See Tables 1 and 2 for detailed description of assigned and fixed parameters used in the models. For the SFM, the intestine was viewed as two tissue subcompartments (serosa and enterocytes) perfused by the serosal (70–95% total intestinal flow) and enterocyte (5–30% total intestinal flow) flow (right panel). This intestinal unit may be substituted into the full PBPK(TM)-PD model to obtain the full PBPK(SFM)-PD model. 192 Ramakrishnan et al.

Cyp27b1 as a subcompartment of the kidney was not considered, because the 2003; Fan et al., 2010). For SFM, the fraction of intestinal blood flow perfusing 21 synthesis rate was low (50 fmol h ) (Hsu et al., 1987) and Cyp27b1 synthesis was the enterocyte region (fQ)is5–30% of the total intestinal blood flow (Cong et al., immediately and completely inhibited upon administration of 1,25(OH)2D3 (Quach 2000; Pang and Chow, 2012). The model suggests that drug in systemic et al., 2015). The mPBPK-PD model (Fig. 1) described baseline concentrations of circulation (e.g., from i.v. dosing) would be partially shunted away from the

1,25(OH)2D3 in tissues (CT,baseline) as functions of CP,baseline (baseline plasma enterocyte region, whereas for by-mouth (PO) dosing, the entire dose would first concentration of 1,25(OH)2D3 or 217 pM) and KT values (or CT/CP,thetissueto reach the enterocytes prior to entering the circulation. In contrast, the TM plasma partition coefficients) (see eq. A4 in Appendix). These values were found suggests that the entire intestinal blood flow perfuses the enterocyte region that is experimentally to be 0.14, 0.34, 0.40, and 0.05 for the liver, kidney, ileum, and brain, indistinguishable from the serosal region. Hence, the SFM suggests the respectively (Chow et al., 2013). The turnover of the Cyp24a1 enzyme was defined occurrence of route-dependent intestinal metabolism (see eqs. A14 and A15 in with a zero-order synthesis rate (kin,Cyp24a1,T) and first-order degradation rate Appendix), and that a greater extent of intestinal metabolism occurs for PO or (kout,Cyp24a1,T). Numerical values of kin and kout are identical inasmuch Cyp24a1baseline i.p. over i.v. dosing (Cong et al., 2000). values are unity. The pharmacodynamic changes [FC of Cyp24a1, or Cyp24a1FC,T (mRNA expression/control expression value)] were assumed to change proportionally Data Fitting and Simulations with the ratio of the relevant tissue 1,25(OH)2D3 concentration (CT)toitsbaseline value (CT/CT,baseline)(seeeq.A7inAppendix). ADAPT5 (version 5, Biomedical Simulations Resource, University of The Full PBPK-PD Model. The full PBPK-PD model incorporated FCs in Southern California, Los Angeles, CA) was used for model fitting and mRNA expression of both Cyp24a1 and Cyp27b1 (Fig. 2). The model is similar simulations. The mPBPK-PD (Fig. 1) and full PBPK-PD (Fig. 2) were used, to the minimal model in most respects except for the definition of the with the intestinal compartment being described the traditional way (TM) or as pharmacodynamics. Induction of Cyp24a1 and inhibition of Cyp27b1 were SFM (for intestinal compartment), with two subcompartments representing the

described using indirect response equations (Dayneka et al., 1993; Sharma and enterocyte and serosal regions (Cong et al., 2000). The initial condition for the Downloaded from

Jusko, 1996; Mager et al., 2003), comprising the full sigmoidal Emax (Imax), EC50 amount of 1,25(OH)2D3 in gut lumen was the i.p. dose administered (eq. A11 in (IC50), and Hill coefficients. Appendix). In contrast, the i.v. dose was administered to plasma compartment Nested TM versus SFM in PBPK-PD Modeling. To account for differ- directly (eq. A6 in Appendix). ences in intestinal blood/plasma flow to the enterocyte region of the intestine, we Fitting. Fitting with equations shown in the Appendix for the minimal and full highlighted the SFM to contrast with the TM for both minimal PBPK and full PBPK-PD models to repeated i.p. data (120 pmol 4) was performed. Rate

PBPK models (right panels of Figs. 1 and 2). Pang and colleagues have viewed equations for plasma flows (QT) and tissue volumes (VT), common physiologic

the intestine as two tissue subcompartments for the SFM, with an enterocyte parameters (Davies and Morris, 1993; Brown et al., 1997) (Table 1), were dmd.aspetjournals.org compartment consisting of absorptive/secretory transporters at the apical assigned for fitting of the PBPK-PD models with nested SFM and TM (see membrane facing the lumen, metabolic enzymes within and a basolateral side Appendix). The initial condition or the baseline plasma concentration of facing the blood, and a serosal compartment acting only as a storage or 1,25(OH)2D3 [CP(0) or CP,baseline] was assigned the measured value (217 pM) distribution compartment (Cong et al., 2000; Doherty and Pang, 2000; Pang, (Chow et al., 2013). The baseline concentration of 1,25(OH)2D3 in tissue

TABLE 1

Assigned physiologic parameter values for PBPK-PD models at ASPET Journals on June 27, 2016

Volume (V) and plasma flow (Q) were obtained from Davies and Morris, 1993, and Brown et al., 1997; KT and plasma baseline concentrations, CP,baseline, were obtained experimentally (Chow et al., 2013).

Parameter (Unit) Definition Value

VP (mL) Plasma volume 0.962 VK (mL) Kidney volume 0.417 VL (mL) Liver volume 1.37 a VI (mL) Intestine volume 0.632 Ven ¼ fQVI (mL) Enterocyte volume — Vser ¼ (1-fQ)VI (mL) Serosal volume — VBr (mL) Brain volume 0.412 Vperi or Vothers (mL) Peripheral compartment volume 18.5 21 b QK (mL min ) Plasma flow to kidney 0.733 21 b QL (mL min ) Plasma flow to liver 1.3 21 b QHA (mL min ) Hepatic arterial plasma flow rate 0.26 21 a,b QI (mL min ) Plasma flow to intestine 1.04 21 Qen¼ fQQI (mL min ) Plasma flow to enterocyte — 21 Qser¼(1-fQ)QI (mL min ) Plasma flow to serosa — 21 b QBr (mL min ) Plasma flow to brain 0.266 21 c QCO (mL min ) Plasma cardiac output 8.04 KL Partition coefficient of liver (liver to plasma concentration ratio) 0.15 KK Partition coefficient of kidney (kidney to plasma concentration 0.34 ratio) KBr Partition coefficient of brain (brain to plasma concentration 0.05 ratio) KI ¼ Kser ¼ Ken Partition coefficient of intestine, serosa, or enterocyte (intestine 0.4 to plasma concentration ratio) CP;baseline (pM) Baseline plasma 1,25(OH)2D3 concentration 217 d 21 CK;baseline (pmol kg ) Baseline renal tissue 1,25(OH)2D3 concentration 73.5 d 21 CL;baseline (pmol kg ) Baseline liver tissue 1,25(OH)2D3 concentration 30.3 d 21 CBr;baseline (pmol kg ) Baseline brain tissue 1,25(OH)2D3 concentration 10.8 d 21 CI;baseline (pmol kg ) Baseline intestinal tissue 1,25(OH)2D3 concentration 86.6

a For TM, fQ = 1; for SFM, fQ was estimated from fitting; fQ VI =Ven and (1-fQ)VI =Vser;fQQI =Qen and (1-fQ)QI =Qser. b Values of tissue plasma flow were calculated as % plasma cardiac output (QCO) (Brown et al., 1997). c 21 Plasma cardiac output (QCO) for mice was calculated using an allometric relationship: QCO,mice (mL min ) = 275 (1-Hct) (body weight of mice in kg)0.75, where Hct is hematocrit (Brown et al., 1997). d CT,baseline was calculated according to CP,baseline and the apparent KT value (CT,baseline/CP,baseline) for kidney, liver, brain, and intestine 0.34, 0.14, 0.05, and 0.4 (Chow et al., 2013); see Appendix. PBPK-PD Modeling of 1,25(OH)2D3 in Mice 193

(CT,baseline) was expressed as a function of CP,baseline, as described in eq. A4 in the For a drug that is confined to the plasma and does not distribute into red blood Appendix. cells, the measured plasma concentration may be converted to blood concentra-

We assumed that 1,25(OH)2D3 is confined to the plasma space and inter-relate tion, from the equality: VB (1-Hct)CP =CBVB. plasma concentrations and plasma flows in the rate equations that also relate blood concentrations and blood flows; we further defined K as the tissue to CB T CP¼ ð1Þ plasma concentration ratio (CT/CP). The assumption that 1,25(OH)2D3 is 1-Hct confined to the plasma space is consistent with compartmental estimates of 21 Plasma volume (VP) and plasma flow (QP) could be expressed in terms of 61.5 mL kg or 1.23 mL for a 20 g mouse for V1, the central volume of blood volume (VB) and blood flow (QB). distribution of 1,25(OH)2D3 for mice given the 120 pmol i.v. dose (Quach et al., 2015). By assuming a hematocrit (Hct) of 0.45, then the estimated blood volume for VP ¼ð1-HctÞVB ð2Þ the central compartment is 1.23/(1-0.45) or 2.24 mL. The value is similar to the ¼ð Þ ð Þ QP 1-Hct QB 3 published value for blood volume for the mouse (1.5-2.5 mL; web.jhu.edu/animalcare/ procedures/mouse.html), suggesting that 1,25(OH)2D3 is indeed confined to the A Hct value of 0.45 was used for mice (average value obtained from Charles plasma. River Laboratories, St. Constance, QC, Canada). Equations in the Appendix were Downloaded from dmd.aspetjournals.org at ASPET Journals on June 27, 2016

Fig. 3. FC in renal Cyp27b1 as well as renal, intestinal, hepatic, and brain Cyp24a1 mRNA expression versus the corresponding tissue 1,25(OH)2D3 concentrations g 1 EmaxCT following the first, second, third, fourth, and all four doses of 120 pmol [data of (Chow et al., 2013)]. Equation 5 for induction Cyp24a1 ; ¼ð1 þ g g Þ and eq. 6 for FC T 1 þ 1 EC50 CT g 2 Imax CK inhibition Cyp27b1 ; ¼ð1- g g Þ were used here, where g and g denote the Hill coefficient (g = 1) of Cyp27b1 and Cyp24a1, respectively, and baseline values of FC K 2 þ 2 1 2 IC50 CK Cyp24a1FC,T and Cyp27b1FC,K =1. 194 Ramakrishnan et al. used for model fitting. The initial 1,25(OH)2D3 baseline concentration Simulations (C ) in tissue was estimated from K and C for both models T,baseline T P,baseline Simulations were performed to examine whether the mPBPK-PD and full Appendix (eq. A4, ). Because a rich data set existed, tissue 1,25(OH)2D3 PBPK-PD models, with TM- or SFM-nested as the intestine compartment, were concentrations were used for fitting, not only for the full PBPK-PD models, able to predict the rebound phenomenon from previous repeated and single i.p. but also for the mPBPK-PD models. For data fitting of both models, a naive- administration (120 pmol) (Chow et al., 2013) as well as assayed data for repeated pooled data analysis approach was used, in which all data were modeled i.v. administration with different (2, 60, and 120 pmol) 1,25(OH)2D3 doses (Quach simultaneously in ADAPT5 using the maximum likelihood estimator. The et al., 2015). Final parameter estimates and assigned constants were used for variance model was defined as: simulations. The prediction errors (PE), defined as PEi =Cpred,i 2 Cobs,i (Sheiner and Beal, 1981; Wu, 1995), were calculated to compare the precision of mPBPK- VAR ¼ðs þ s Yðu; t ÞÞ2 ð4Þ i 1 2 i PD and full PBPK-PD model when predicting single i.p. and repeated i.v. data. th The median prediction error (MPE) and median absolute prediction error (MAPE) with s1 and s2 as the variance model parameters, and Yðu; tiÞ as the i predicted value from the pharmacokinetic model. Different variance pa- were used to estimate accuracy and precision, respectively: n o rameters were used for 1,25(OH)2D3 concentrations in plasma/tissues and ¼ ; ; ð Þ the mRNA expression of VDR genes (Cyp27b1 and Cyp24a1). The final MPE Median Cpred i-Cobs i 0-8 days 7 model was selected based on goodness-of-fit criteria, which included model convergence, parameter precision, and visual inspection of predicted where MPE is the median value of the prediction error from time 0 to 8 days. versus observed values and residual plots. The F test was used to com- n o MAPE ¼ Median C ; -C ; ð8Þ pare goodness of fit of the nested TM and SFM models (Boxenbaum pred i obs i 0-8 days et al., 1974).

Downloaded from The mPBPK-PD Model. For fitting of the mPBPK-PD model, Cyp27b1FC,K where MAPE is the median value of the absolute PE, Cpred;i-Cobs;i , from time was omitted in the kidney compartment due to the low synthesis rate of the 0 to 8 days. The Wilcoxon matched pair test was used to compare MPE and enzyme (see eqs. A6 - A16, Appendix). The FC of mRNA expression of MAPE between different models for the same set of data (Wu, 1995). Cyp24a1, or Cyp24a1FC,T, was expressed as a linear function of CT/CT,baseline (see eq. A7, Appendix). Quantitative Real-Time Polymerase Chain Reaction The Full PBPK-PD Model. For full PBPK-PD model fitting, the indirect To evaluate the relative expressions of Cyp24a1 in tissue, total RNA, obtained response models with E (or I )andEC (or IC ) values were used for max max 50 50 from kidney, ileum, liver, and brain samples from control mice, was extracted dmd.aspetjournals.org description of induction of Cyp24a1 and inhibition of Cyp27b1, respectively using the TRIzol extraction method (Sigma-Aldrich, Mississauga, ON, Canada), (see eqs. A18 and A19, Appendix). These constants were estimated by according to manufacturer’s protocol with modifications (Chow et al., 2011). regression of the FC of the enzymes against the relevant tissue 1,25(OH) D 2 3 cDNA (total of 1.5 mg) was synthesized from RNA using the high capacity concentration (Fig. 3). cDNA reverse transcription kit (Applied Biosystems by Life Technologies, The Cyp24a1 in tissue was expressed as: FC,T Burlington, ON, Canada), followed by quantitative real-time polymerase chain

g1 reaction using the SYBR Green detection system. mRNA data were normalized ¼ð þ EmaxCT ÞðÞ Cyp24a1FC;T 1 g g 5 to cyclophilin for calculation of the relative change in gene expression in terms of 1 þ 1

EC C at ASPET Journals on June 27, 2016 50 T FC (Chow et al., 2011). with Emax as the maximum stimulatory effect, EC50 as the tissue 1,25(OH)2D3 concentration producing 50% of Emax, and g1 as the Hill coefficient of Cyp24a1 Results in tissue (Mager et al., 2003; Quach et al., 2015). Estimation of Imax,IC50,Emax, and EC50 The Cyp27b1FC,K in the kidney was expressed as: Plots of FCs of Cyp24a1 and Cyp27b1 mRNA expression levels in g2 ¼ð ImaxCK ÞðÞ kidney, liver, ileum, and brain versus tissue 1,25(OH)2D3 concentration Cyp27b1FC 1- g g 6 2 þ 2 IC50 CK are shown in Fig. 3. The Imax (and IC50) and Emax (and EC50) values were obtained after regression of Cyp27b1FC,K and Cyp24a1FC,T with Imax as the maximum inhibitory effect, IC50 as the renal 1,25(OH)2D3 expression against the corresponding tissue 1,25(OH)2D3 concentra- g concentration producing 50% of Imax, and 2 as the Hill coefficient of Cyp27b1 tions with eqs. 5 and 6. Parameters for inhibition of Cyp27b1 and in kidney (Mager et al., 2003; Quach et al., 2015). Regression of data from each dose and for the entire data set was performed according to eqs. 5 and 6. Because induction of Cyp24a1 mRNA expression for each dose and for the there were no trends among these estimates, final estimates were obtained from combined doses are summarized in Table 2. Similar values of EC50 and E for the induction and IC and I for the inhibition were the fit to all data. Values of Emax and Imax were assigned based on our final max 50 max estimates, whereas the EC50 and IC50 estimates were used as initial estimates, obtained among the first, second, third, and fourth doses. Finally, the and final values were obtained from model fitting. EC50 and IC50 values for the composite fit (combined doses) were used

TABLE 2

Emax/Imax and EC50/IC50 for concentrations versus response (FC) curves following repeated i.p. administration

Fitted Valuesb 1st Dosea 2nd Dosea 3rd Dosea 4th Dosea Combined Dosesa Fitted Valuesb PBPK(SFM)-PD PBPK(TM)-PD

Imax IC50 Imax IC50 Imax IC50 Imax IC50 Imax IC50 IC50 IC50

Cyp27b1FC;K 1 184 0.8 230 0.16 350 0.75 260 1 160 77.3 (7.03) 66.8 (10.5)

Emax EC50 Emax EC50 Emax EC50 Emax EC50 Emax EC50 EC50 EC50

Cyp24a1FC;K 72 170 72 200 68 250 66 130 65 260 113 (1.75) 248 (1.87) Cyp24a1FC;I 500 727 650 1700 680 2600 630 2850 675 2300 2640 (0.703) 2270 (2.89) —— Cyp24a1FC;L 34 2250 47.7 3040 450 2530 170 2270 2600 (0.319) 2900 (0.146) Cyp24a1FC;Br 49 250 49 235 36 119 43 74 49 106 235 (1.15) 125 (8.35)

a The optimized parameters used to simulate correlation between tissue 1,25(OH)2D3 and FC of Cyp24a1 and Cyp27b1. bFitted values [parameter estimate (CV%)] were obtained after fitting the data with the full PBPK-PD models (see other fitted parameters in Table 3). PBPK-PD Modeling of 1,25(OH)2D3 in Mice 195 as initial estimates in model fitting, and the Emax and Imax values between the fits to both data sets. Fitted values for EC50 were much were assigned. The Emax values estimated in this study are about lower than those obtained by Quach et al. (2015), who used plasma 61–70% of those estimated by Quach et al. (2015). The estimated instead of tissue 1,25(OH)2D3 concentrations for fitting. Parame- Imax was slightly lower than that by Quach et al. (2015), but the IC50 ters estimated from fitting with tissue 1,25(OH)2D3 concentrations estimate was similar. Overall, there was good correspondence are more appropriate. Downloaded from dmd.aspetjournals.org at ASPET Journals on June 27, 2016

Fig. 4. Observed (closed circles) and fitted (spline lines) concentration-time profiles of 1,25(OH)2D3 and time course of FC of tissue Cyp24a1 and Cyp27b1 mRNA after multiple i.p. doses [data of (Chow et al., 2013)] using mPBPK-PD model with nested TM (dashed lines) and SFM (solid lines). 196 Ramakrishnan et al. Downloaded from dmd.aspetjournals.org at ASPET Journals on June 27, 2016

Fig. 5. Observed (closed circles) and fitted (spline lines) concentration-time profiles of 1,25(OH)2D3 and time course of FC of tissue Cyp24a1 and Cyp27b1 mRNA after multiple i.p. doses [data of (Chow et al., 2013)] using the full PBPK-PD model with nested TM (dashed lines) and SFM (solid lines).

Fitting to Minimal and Full PBPK-PD Models naive-pooled 1,25(OH)2D3 concentrations and Cyp24a1FC,T mRNA The observed and model-fitted concentration-time profiles using the expression for the multiple i.p. doses well. These models predicted minimal and full PBPK-PD models with nested TM and SFM are similar trends for the plasma and tissue 1,25(OH)2D3 profiles and shown in Figs. 4 and 5, respectively. All of the models characterized the showed that the peak concentration (Cmax) was reached within PBPK-PD Modeling of 1,25(OH)2D3 in Mice 197

TABLE 3 Fitted parameters [estimate and (CV%)] obtained from mPBPK-PD and full PBPK-PD models with TM and SFM nested within the models

Fitted Parameters Definition mPBPK(TM)-PD mPBPK(SFM)-PD Full PBPK(TM)-PD Full PBPK(SFM)-PD

fQ Fraction of intestinal flow to 1 0.114 (2.26) 1 0.105 (0.533) enterocyte region fd Fraction of cardiac output to 0.0021 (0.021) 0.0032 (0.025) 0.0035 (0.002) 0.0048 (0.001) peripheral compartment 21 ka (h ) Absorption rate constant of 1.43 (0.011) 1.26 (0.013) 1.50 (0.004) 1.61 (0.0021) 1,25(OH)2D3 21 kdeg (h ) Degradation rate constant of 0.0042 (6.06) 0.0013 (11.6) 0.0017 (2.34) 0.0012 (6.65) 1,25(OH)2D3 in lumen Kperi or Kothers Partition coefficient of peripheral/ 0.117 (0.025) 0.161 (0.032) 0.272 (0.003) 0.335 (0.003) other compartment 21 Rsyn (fmol h ) Endogenous synthesis rate of 50.9 (0.26) 23.0 (0.238) 31.0 (0.247) 21.5 (0.116) 1,25(OH)2D3 —— kin;Cyp27b1;K or Turnover rate constants of renal 0.220 (11.5) 0.245 (9.72) 21 kout;Cyp27b1;K (h ) Cyp27b1 —— g2 Hill coefficient for indirect response 3.57 (14.0) 2.71 (12.1) of renal Cyp27b1 function 21 a fLCLint;met;L (mL h ) Hepatic metabolic intrinsic clearance 0.031 (0.841) 0.0782(2.82) 0.0043 (1.78) 0.0010 (3.03) of 1,25(OH)2D3 via hepatic Cyp24a1 Downloaded from 21 a fICLint;met;I (mL h ) Intestinal metabolic intrinsic 0.229 (0.348) 0.220 (0.196) 0.0011 (0.038) 0.0014 (0.412) clearance of 1,25(OH)2D3 via intestinal Cyp24a1 21 a fKCLint;met;K (mL h ) Renal metabolic intrinsic clearance of 0.053 (0.435) 0.0641 (0.782) 0.0242 (0.008) 0.0280 (0.0066) 1,25(OH)2D3 via renal Cyp24a1 21 a fBrCLint;met;Br (mL h ) Brain metabolic intrinsic clearance of 0.030 (15.0) 0.0345 (3.37) 0.0006 (8.30) 0.0003 (0.412) 1,25(OH)2D3 via brain Cyp24a1 L —— dmd.aspetjournals.org g1 Hill coefficient for indirect response 1.24 (2.45) 1.75 (0.020) function of hepatic Cyp24a1 K —— g1 Hill coefficient for indirect response 2.64 (0.665) 3.59 (0.081) function of renal Cyp24a1 I —— gI Hill coefficient for indirect response 0.985 (1.22) 2.09 (0.002) function of intestinal Cyp24a1 Br —— g1 Hill coefficient for indirect response 0.878 (0.148) 0.576 (0.076) function of brain Cyp24a1 kin;Cyp24a1;L or Turnover rate constant of hepatic 0.0413 (0.052) 0.044 (0.708) 0.045 (0.006) 0.047 (0.008) 21 kout;Cyp24a1;L (h ) Cyp24a1 at ASPET Journals on June 27, 2016 kin;Cyp24a1;I or Turnover rate constant of intestinal 0.218 (0.278) 0.174 (0.485) 0.489 (0.005) 0.287 (0.067) 21 kout;Cyp24a1;I (h ) Cyp24a1 kin;Cyp24a1;K or Turnover rate constant of renal 0.0256 (0.038) 0.0238 (0.026) 0.012 (0.065) 0.047 (0.008) 21 kout;Cyp24a1;K (h ) Cyp24a1 kin;Cyp24a1;Br or Turnover rate constant of brain 1.20 (2.73) 2.10 (4.52) 1.28 (0.051) 1.87 (5.10) 21 kout;Cyp24a1;Br (h ) Cyp24a1 Cyp24a1;L —— SM1 Power coefficient on renal Cyp24a1 0.931 (0.112) 0.766 (0.933) enzyme turnover Cyp24a1;K —— SM2 Power coefficient on hepatic Cyp24a1 1.27 (0.120) 1.33 (0.056) enzyme turnover Cyp24a1;Br —— SM3 Power coefficient on brain Cyp24a1 0.804 (0.389) 0.800 (0.498) enzyme turnover Cyp24a1;I —— SM4 Power coefficient on intestinal 1.40 (0.116) 1.42 (0.160) Cyp24a1 enzyme turnover HK Power coefficient on renal intrinsic 0.828 (0.0622) 0.784 (0.0372) —— clearance HL Power coefficient on hepatic intrinsic 1.86 (0.0769) 0.470 (2.23) —— clearance HBr Power coefficient on brain intrinsic 0.142 (17.8) 0.374 (5.68) —— clearance HI Power coefficient on intestinal 0.230 (0.246) 0.054 (7.82) —— intrinsic clearance AIC Akaike information criteria 9367 9270 10,006 9993 WSSR Weighted sum of squared residuals 1684 1644 1664 1478 df b Degrees of freedom 989 988 986 985 (critical F = 3.84) Calculated F 24.0d versus mPBPK 5.93d versus mPBPK 55.3d versus mPBPK score (TM)-PD (TM)-PD (SFM)-PD F valuec 124d versus PBPK (TM)-PD

a Fitted CLint,met,T values were adjusted by corresponding unbound fraction of 1,25(OH)2D3 in tissue (fT). bDegrees of freedom are the number of data points (n = 1010) used in the model minus the number of parameters being fitted. WSS 2 WSS cF score was calculated using j i dfi , where df . df . WSSi dfj 2 dfi j i dF score suggests a significant improvement in goodness of fit (critical F = 3.84) (Boxenbaum et al., 1974); the rank order of goodness of fit (from poorest to best for optimal fit): mPBPK(TM)-PD , mPBPK(SFM)-PD , full PBPK(TM)-PD , full PBPK(SFM)-PD model. 198 Ramakrishnan et al.

30 minutes after dosing, and adequately described that the concentration concentrations, better matched the observed 1,25(OH)2D3 dipped below the baseline about 24 hours postdose. Furthermore, both concentration-time profiles compared with those from the full PBPK mPBPK-PD and full PBPK-PD models were able to describe the (TM)-PD model (Fig. 5). Values from the F test attests that the full induction pattern of Cyp24a1. The final parameters were estimated with PBPK(SFM)-PD is the better model (Table 3). good precision (low percentage of coefficient of variation, or coefficient Validation of the Models. To validate the robustness of the models, of variation) and are summarized in Tables 2 and 3. Both models simulations were performed for the repetitive i.p. dosing of 120 pmol estimated a similar, quick absorption phase, with ka values of about (Chow et al., 2013) and escalating i.v. doses (2, 60, and 120 pmol) 21 1.2–1.6 h , and a slow luminal degradation (kdeg) phase. Values Quach et al., 2015). Simulations described well the plasma and tissue estimated for the net synthesis rate (Rsyn), for 1,25(OH)2D3 were 21–51 concentrations rising and then falling below baseline after 24–48 hours fmol h21, and were in agreement with those (50 and 81.6 fmol h21) and gradually returning back to basal level following repetitive (Figs. 7 previously observed by Hsu et al. (1987) and Quach et al. (2015). and 8) or single (Figs. 9 and 10) i.p. administration (120 pmol). Both Other fitted parameters that relate to 1,25(OH)2D3 metabolism and PBPK-PD models also predicted the Cyp24a1FC,T and Cyp27b1FC,K enzyme turnover (e.g., fTCLint,met,T,kin and kout for Cyp27b1 and mRNA levels in tissue returning to basal level (=1), suggesting that the Cyp24a1) were, however, different between the minimal and full up- and downregulation of VDR target genes disappeared within 10–14 PBPK-PD models. Upon checking the relative mRNA expressions of days after discontinuation of treatment (Figs. 7–10). Cyp24a1 among these tissues, a rank order may be established, as The robustness of the model for prediction of escalating i.v. doses (2, follows: kidney . intestine . liver . brain, varying from 1.85 for 60, and 120 pmol every other day for 6 days; Figs. 11 and 12) was also kidney to 1.0 for brain (Fig. 6). The rank order was similar to the pattern examined. All models were able to predict the pharmacokinetics of Downloaded from for fitted fTCLint,met,T values for Cyp24a1-mediated metabolism 1,25(OH)2D3 following repeated i.v. administration, although data for (kidney . ileum . liver .. brain) for the full PBPK(SFM)-PD. the low doses were not as well predicted by the mPBPK-PD models. The rank order for other models, however, did not follow the same After comparison of prediction errors (P , .05; Table 4), the full PBPK- trend. Relative protein levels, although assayed, were not comparable PD model was found to be more consistent with data and more robust among tissues because the housekeeping gene also differed (data not than the mPBPK-PD model, as defined by the MPE and MAPE values shown). Therefore, these values were not used for comparison. Kperi in Table 4. Moreover, both observed and predicted Cmax in ileum dmd.aspetjournals.org values varied between 0.1 and 0.3 and were in line with tissue to plasma following i.v. administration were unexpectedly lower than the Cmax partition coefficients of 1,25(OH)2D3 in other tissues (Table 1). It was after i.p. administration (Fig. 12 versus Fig. 5) despite the bioavailabil- noted, however, that there were differences in the model-specific ity of 0.84 6 0.16 [from the dose-corrected area under the curve ratios scaling factors or coefficients, as follows: for tissue-specific power of i.p./i.v., (AUCi.p./Dosei.p.)/(AUCi.v./Dosei.v.) for all 4 models], coefficient in tissue (0.8–1.4) that scales the pharmacodynamic suggesting a lesser distribution of 1,25(OH)2D3 into the enterocyte after equations for the mPBPK-PD model; for power coefficient in tissue i.v. dosing, a flow pattern lending support to the SFM model. The full (0.05–1.9) that scales the power of FC of Cyp24a1; and for the Hill PBPK(SFM)-PD model was better for describing the route-dependent coefficients g1 and g2 (0.6–3) that relate the FC in the Hill equation. kinetics of 1,25(OH)2D3 for i.p. dosing than other types of PBPK-PD at ASPET Journals on June 27, 2016 Other fitted parameters, such as fTCLint,met,T for metabolism and models as it captured the lower Cmax in ileum following i.v. adminis- kin,enzyme,T (or kout,enzyme,T) for enzyme turnover rates, varied between tration (Fig. 11 versus Fig. 12). models, probably due to differences in the way of defining the scaling factor and FC of enzyme. Comparison of TM and SFM. Pictorially, both the TM- and SFM- Discussion nested in the mPBPK-PD and full PBPK-PD models were able to Vitamin D synthesis and disposition is a complex, multistage process predict the data equally well (Figs. 4 and 5). However, the F critical occurring in different tissues and is tightly regulated. Hence, we values distinguished the subtle improvement afforded by the models. revisited the rich i.p. and i.v. data of 1,25(OH)2D3 in mice to gain a The results showed that the fit with the SFM was superior compared more physiologically relevant perspective. The modeling design was with the TM in both the minimal and full PBPK-PD models. The fitted based on measurements of 1,25(OH)2D3 concentrations and regulatory value of fQ was ;11% for the SFM, and the presence of segregated flow enzymes that are under feedback control by 1,25(OH)2D3-bound VDR in the intestine significantly improved the goodness of fit (Table 3). The over a sufficiently long duration following multiple i.p. and i.v. doses fitted fQ value of about 11% is consistent with published values of (Chow et al., 2013; Quach et al., 2015). Moreover, a dense sampling morphine, digoxin, and benzoic acid (Pang and Chow, 2012), and frequency had been adopted previously to provide rich temporal deviates from the value of 1 projected by the TM. Predictions from the profiles that best capture the disposition of 1,25(OH)2D3 in plasma full PBPK(SFM)-PD model, especially for lower plasma and tissue and tissues. In addition, the dynamics of critical enzymes involved in the metabolism of 1,25(OH)2D3 were measured in parallel. Quach et al. (2015) employed compartmental pharmacokinetic/ pharmacodynamic (PK/PD) modeling for i.v. 1,25(OH)2D3 data obtained from escalating doses to relate 1,25(OH)2D3 kinetics and the feedback inhibition of Cyp27b1 and induction of Cyp24a1. Finer mechanistic details can be obtained by applying PBPK models that are parameterized based on physiologic system components, functions, and tissue compartments that are connected by plasma or blood flow rates. The minimal PBPK approach is proven to be superior over compartmental models and provides a greater mechanistic insight and more interpretable and physiologically meaningful pharmacokinetic parameters, especially when only plasma or blood data are available (Cao and Fig. 6. Baseline levels of relative mRNA expression for Cyp24a1 in kidney, ileum, Jusko, 2012). It adopts Fick’s law of perfusion by incorporating liver, and brain in control mice. fractional distribution (fd) to account for organ/tissue lumping and PBPK-PD Modeling of 1,25(OH)2D3 in Mice 199 Downloaded from dmd.aspetjournals.org Fig. 7. Simulated (lines) concentration-time profiles of 1,25(OH)2D3 and time course of FC of tissue Cyp24a1 and Cyp27b1 mRNA expression following repeated i.p. doses [data of (Chow et al., 2013)] over 30 days to show the rebound phenomenon. Data were simulated using mPBPK-PD models with nested TM (dashed lines) and SFM (solid lines) described in the Appendix and parameters from Tables 2 and 3. at ASPET Journals on June 27, 2016

variability in cardiac output (QCO), and has been successfully applied to complexity of these types of developed PBPK models are conveniently describe the kinetics of small molecules as well as disposition pathways reduced using lumping approaches, in which tissues with similar and sites of elimination of monoclonal antibodies (Cao et al., 2013; kinetics are grouped together to provide a simpler approach with fewer Li et al., 2014), although additional parameters, including vascular compartments than a whole-body PBPK model incorporating all tissues reflection coefficients, may be needed. The dimensionality and and organs (Nestorov et al., 1998; Pilari and Huisinga, 2010). 200 Ramakrishnan et al. Downloaded from dmd.aspetjournals.org at ASPET Journals on June 27, 2016

Fig. 8. Simulated (lines) concentration-time profiles of 1,25(OH)2D3 and time course of FC of tissue Cyp24a1 and Cyp27b1 mRNA expression following repeated i.p. doses [data of (Chow et al., 2013)] over 30 days to show the rebound phenomenon. Data were simulated using the full PBPK-PD models with nested TM (dashed lines) and SFM (solid lines) described in the Appendix and parameters from Tables 2 and 3.

We first adopted a minimal PBPK-PD model to describe the with tissue 1,25(OH)2D3 concentrations (eq. A7, Appendix). The FC of pharmacokinetics and pharmacodynamics. The mPBPK-PD models pro- enzymes are related to CT/CT,baseline linearly without any knowledge of vide a more simplistic framework and relate Cyp24a1 relative expression Emax/EC50 and Imax/IC50 values. The model was further simplified by PBPK-PD Modeling of 1,25(OH)2D3 in Mice 201 Downloaded from dmd.aspetjournals.org

Fig. 9. Observed (closed circles) versus simulated (lines) concentration-time profiles of 1,25(OH)2D3 and the rebound to baseline levels after a single i.p. dose for the TM (dashed lines) and SFM (solid lines) nested within mPBPK-PD models [data of (Chow et al., 2013)] to show the rebound phenomenon. at ASPET Journals on June 27, 2016

assuming the Cyp27b1 effects on Rsyn as minimal or negligible. The mPBPK-PD model that correlated enzyme expressions with baseline mPBPK-PD model adequately described the 1,25(OH)2D3 and 1,25(OH)2D3 concentrations via simplified, linear relationshipsislessable Cyp24a1FC,T (Figs. 4, 7, and 9). The final PK/PD parameters were to adequately describe the concentration-dependent pharmacodynamic estimated with good precision (low coefficient of variation) (Table 3). This behavior of 1,25(OH)2D3 (Figs. 11 and 12; Table 4). simplified approach is particularly useful when pharmacodynamic data are We also employed the full PBPK-PD models that use indirect scarce and Emax/Imax or EC50/IC50 values are unavailable. However, the response equations (eqs. A18 and 19, Appendix) for describing the 202 Ramakrishnan et al. Downloaded from dmd.aspetjournals.org at ASPET Journals on June 27, 2016

Fig. 10. Observed (closed circles) versus simulated (lines) concentration-time profiles of 1,25(OH)2D3 and the rebound to baseline levels after a single i.p. dose for the TM (dashed lines) and SFM (solid lines) nested within the full PBPK-PD models [data of (Chow et al., 2013)] to show the rebound phenomenon.

pharmacokinetics and pharmacodynamics of 1,25(OH)2D3. The full providing a more complete description of the complex kinetics of PBPK-PD model requires dense sampling in multiple tissues and a 1,25(OH)2D3. Expectedly, inclusion of the inhibitory effect of diverse array of assigned parameters, including Cyp27b1 for the 1,25(OH)2D3 on Cyp27b1-mediated endogenous synthesis (Rsyn) synthesis of 1,25(OH)2D3 and Cyp24a1 for catabolism, thereby of 1,25(OH)2D3 in kidney, which normally accounts for the bulk PBPK-PD Modeling of 1,25(OH)2D3 in Mice 203 Downloaded from dmd.aspetjournals.org at ASPET Journals on June 27, 2016

Fig. 11. Observed (closed circles) versus simulated (spline lines) concentration-time profiles of 1,25(OH)2D3 and time course of FC of tissue Cyp24a1 and Cyp27b1 mRNA after multiple i.v. doses (given every 2 days for 6 days) in plasma, kidney, liver, ileum, and brain using mPBPK-PD models with nested TM (dashed lines) and SFM (solid lines) for describing the intestine compartment [data of (Quach et al., 2015)]. Data were simulated with PBPK-PD using inhibition and induction functions described in the Appendix and parameters in Tables 2 and 3. of circulating 1,25(OH)2D3 (Bell, 1998), aptly described the characterizing repeated i.v. dose data ranging from low to higher 1,25(OH)2D3 profiles, the Cyp24a1 and Cyp27b1 expression, as doses(2versus60and120pmol).Thisisexpectedbecause well as the rebound phenomenon well (Figs. 5, 8, and 10). The full Cyp24a1FC,T and Cyp27b1FC,K are related nonlinearly by saturable PBPK-PD model was found to be superior compared with the sigmoidal Emax/Imax equations, providing a more accurate account of mPBPK-PD model in terms of precision according to predic- concentration-dependent pharmacodynamics (shown in eqs. A18 tion errors and prediction accuracy (Table 4), especially when and A19 in the Appendix). 204 Ramakrishnan et al. Downloaded from dmd.aspetjournals.org at ASPET Journals on June 27, 2016

Fig. 12. Observed (closed circles) versus simulated (spline lines) concentration-time profiles of 1,25(OH)2D3 and time course of FC of tissue Cyp24a1 and Cyp27b1 mRNA after multiple i.v. doses (given every 2 days for 6 days) in plasma, kidney, liver, ileum, and brain using the full PBPK-PD models with nested TM (dashed lines) and SFM (solid lines) for describing the intestine compartment [data of (Quach et al., 2015)]. Data were simulated with PBPK-PD using inhibition and induction functions described in the Appendix, and parameters in Tables 2 and 3. In particular, Cmax 1,25(OH)2D3 values in ileum were much overestimated for the TM. The extents of overestimation were less for the full PBPK(SFM)-PD and mPBPK(SFM)-PD models (Fig. 11), with the full PBPK(SFM)-PD model being the best.

The SFM (versus the TM) of the intestinal compartment defines the the peritoneal cavity serving as a reservoir for i.p. dosing, passive, nonvascular (oral or i.p.) route of administration in more physiologi- nonsaturable, and continuous absorption of 1,25(OH)2D3 into the cally meaningful terms and distinguishes the differences in plasma/ enterocyte compartment ensues (Hollander et al., 1978). The SFM was blood flow, transporter, channel, and metabolic enzyme density in the shown to be superior over the TM when these intestinal models enterocyte and serosal regions of the intestine (Cong et al., 2000). With are nested in the mPBPK-PD and full PBPK-PD models. With PBPK-PD Modeling of 1,25(OH)2D3 in Mice 205

TABLE 4 Median prediction error (MPE) and median absolute prediction error (MAPE) calculated using mPBPK-PD and full PBPK-PD models with nested TM and SFM following repeated i.v. administration (q2d 3)

Model Types

TM SFM

mPBPK(TM)-PD Full PBPK(TM)-PD P Valuesa mPBPK(SFM)-PD Full PBPK(SFM)-PD P Valuesa Tissue and i.v. doses (pmol) MPE MAPE MPE MAPE MPE MAPE MPE MAPE MPE MAPE MPE MAPE Plasma 2 365 365 359 359 ,.001b .001b 273 323 103 187 ,.001b .007b Plasma 60 6314 6314 5659 5659 .062 .062 4512 4512 2946 2946 .008b .008b Plasma 120 1398 3991 339 4712 .100 .808 589 5763 57 4233 .010b .040b Kidney 120 20 113 17.5 17.5 .314 .028b 23.52 66.5 4.60 13 .753 .028b Liver 120 114 114 243.4 43.4 .075 .046b 66.4 66.4 2.70 48.1 .028b .028b Ileum 120 667 667 365 365 .028b .028b 374 374 7.70 132 .028b .046b

aThe Wilcoxon matched pair test was performed to compare MPE and MAPE between the mPBPK-PD and full PBPK-PD models. bDifferences between the mPBPK-PD and full PBPK-PD model (P , .05). dC ; C ; 1,25(OH)2D3 given i.p., part of the dose must traverse the enterocyte T baseline ¼ T baseline ð Þ VT QT CP;baseline- A1 layer before reaching systemic circulation, whereas with i.v. adminis- dt KT Downloaded from tration, the entire dose directly enters the circulation. Consequently, more 1,25(OH)2D3 is available for intestinal metabolism following i.p. where CP,baseline,CT,baseline,VT, and QT are the baseline plasma and or PO than i.v. administration (Cong et al., 2000), allowing a greater tissue concentrations, tissue volume, and plasma flow rate through the extent of intestinal metabolism (see Figs. 1 and 2). For 1,25(OH)2D3, tissue compartment, respectively. At steady state, KT, the tissue to however, the extent of first-pass intestinal removal is small because the plasma partition coefficient, is given by the ratio of the tissue to plasma

total clearance is low (Quach et al., 2015). The distinct presence of a concentration: dmd.aspetjournals.org serosal storage compartment better defines the distribution space in the CT;baseline intestine (Cong et al., 2000). The data support the SFM as the preferred ¼ KT ðA2Þ C ; model over the TM, inferring that there is intestinal route-dependent P baseline metabolism, namely, a drug given systemically will be less extracted by For an eliminating tissue, the rate of change in the tissue at the basal the intestine due to the low blood flow rate perfusing the enterocyte level, in absence of induction of the degradating enzyme, Cyp24a1, is region (Cong et al., 2000; Pang, 2003; Pang and Chow, 2012). given by: In conclusion, a biologically plausible model has been developed for at ASPET Journals on June 27, 2016 the quantitative characterization of the roles of Cyp24a1 and Cyp27b1 dCT;baseline ¼ ð CT;baseline Þ VT QT CP;baseline- -fTCT;baselineCLint;met;T in regulating the complex pharmacokinetics of 1,25(OH)2D3 in mice. dt KT Based on the study design of the present study, PBPK-PD modeling ðA3Þ provides a mechanism-based framework for discerning the tissue- specific disposition characteristics of 1,25(OH)2D3. The pharmacody- where fT is the unbound fraction and CLint,met,T is the intrinsic namic effects of 1,25(OH)2D3 are tightly regulated by the endogenous metabolic clearance in the tissue compartment. At steady state, tissue concentrations, and the current models provide various platforms to integrate the absorption, distribution, and metabolism/excretion of QTCP;baseline CT;baseline ¼ ðA4Þ 1,25(OH) D to biologic effects observed in preclinical studies. We QT 2 3 þ fTCLint;met;T demonstrated good utility of the mPBPK-PD as a rational and simplistic KT approach for describing PK/PD interplay when limited pharmacodynamics data are available. The full PBPK-PD model, however, is and superior to mPBPK-PD for describing dose-dependent kinetics and CT;baseline QT 1 K ; ¼ ¼ ¼ utilizes dynamic constants such as Emax,EC50,Imax, and IC50, and T app CP;baseline QT þ 1 þ fTCLint;met;T lastly, the nested SFM better characterizes the route-dependent in- fTCLint;met;T KT KT Q testinal removal/distribution as compared with the TM. The pharma- T K cokinetic models developed in this study could be extended to ¼ T ðA5Þ KTfTCLint;met;T understand the pharmacodynamic regulation of 1,25(OH)2D3 as well 1 þ as other endogenous compounds in a quantitative manner. The models QT may also find utility in predicting 1,25(OH) D disposition for C ; 2 3 As shown in eq. A5, T baseline for eliminating organs yields only the interspecies scaling and for exploration of alternative dosing schemes CP;baseline apparent K (K ) and underestimates the true K (Chen and Gross, and routes of administration to describe the dynamics of 1,25(OH) D T T,app T 2 3 1979), in view of the fact that there is elimination within tissues with in its new therapeutic roles. Cyp24a1. These KT values, however, will not deviate much from the true values because the clearance of 1,25(OH)2D3 is low (Quach et al., 2015), Appendix and the error in underestimation of KT,basedonCT,baseline/CP,baseline, will be small. We further assumed that the KT would stay constant under conditions when Cyp27b1 and Cyp24a1 are altered, and equal KT under Definition of Q, V, KT, and CT,baseline basal conditions. For a noneliminating tissue, the rate of change in the tissue under In the equations to follow, Qperi,QBr,QK,QI, and QL are the plasma basal conditions is given by: flow rates to the peripheral, brain, kidney, intestine, and liver 206 Ramakrishnan et al.

dCK ¼ þ compartments, respectively; Vperi,VBr,VK,VI, and VL are the corre- For the Rate of Change in Kidney. VK dt Rsyn QK CK HK sponding tissue volumes, respectively. Q and V are the plasma CP- fKCKCLint;met;KðCyp24a1 ; Þ , where HK is the power HA P KK FC K arterial flow rate and plasma volume, respectively. coefficient for the kidney and Rsyn is the zero-order, net synthesis rate of endogenous 1,25(OH)2D3 in the kidney. The mPBPK(TM)-PD Model (Fig. 1) þ ð Þ¼ ¼ QKCP;baseline Rsyn ¼ For the Rate of Change of 1,25(OH)2D3 in Plasma. CK 0 CK;baseline since Cyp24a1FC;K 1 QK þ f CL ; ; K K int met K dCP ¼ þ þ þ þ CL þ CK þ CBr K VP - QL QK QBr Qperi CP QL QK QBr ð Þ dt KL KK KBr A9

dCI Cperi For the Rate of Change in Intestine. VI ¼ kaAlumenþ þ Q ; CPð0Þ¼CP;baseline ¼ 217 pM dt peri CI HI Kperi Q CI- fICICLint;met;IðCyp24a1 ; Þ , where HI is the power I KI FC I ðA6Þ coefficient for the intestine.

Q is the plasma flow rate for the peripheral compartment (where peri ð Þ¼ ¼ QICP;baseline ¼ ð Þ Q =f Q ) and is expressed as a fraction (f ) of plasma cardiac CI 0 CI;baseline since Cyp24a1FC;I 1 A10 peri d CO d QI þ f CL ; ; output (QCO), and QL =(QI +QHA) is the total hepatic plasma flow rate, I int met I KI where QHA and QI are the plasma flow of the hepatic artery and portal Downloaded from vein, respectively. For the Rate of Change in Gut Lumen. Changes in 1,25(OH)2D3 concentrations in tissues must first consider the dAlumen rate of change of the degradation enzyme, Cyp24a1. In absence of enzyme ¼ - k þ k A ; A ð0Þ¼Dose : : ðA11Þ dt a deg lumen lumen i p dCyp24a1T;baseline ¼ induction, dt kin;Cyp24a1;T-kout;Cyp24a1;TCyp24a1T;baseline,where (kin,Cyp24a1,T) is the zero-order production rate constant and (kout,Cyp24a1,T) Here, absorption of the i.p. dose is assumed to occur solely from the is the first-order degradation rate constant of the enzyme. Values of gut lumen, and is parameterized by the first-order absorption (ka) and dmd.aspetjournals.org ¼ degradation (k ) rate constants in the gut lumen; A is the amount kin,Cyp24a1,T and kout,Cyp24a1,T are identical when Cyp24a1T;baseline 1. deg lumen For mPBPK-PD model, Cyp24a1FC,T was assumed to change of 1,25(OH)2D3 in the gut lumen. dCL ¼ þ CI directly with the relevant tissue 1,25(OH)2D3 concentration (CT), For the Rate of Change in Liver. VL dt QHACP QIK C HL I Q L f C CL ; ; ðCyp24a1 ; Þ , where HL is the power co- which in turn is expressed relative to the corresponding baseline LKL L L int met L FC L concentrations ( CT ). With induction of Cyp24a1, k is efficient for the liver. ( ) CT;baseline in,Cyp24a1,T increased by the factor, ð CT ÞSMT where SM is the tissue-specific C ; T QI T baseline þ Q CP;baseline fICLint;met;I HA power coefficient. þ at ASPET Journals on June 27, 2016 1 Q dCyp24a1T ¼ ð CT ÞSMT ð Þ¼ ¼ I kin;Cyp24a1;T -kout;Cyp24a1;TCyp24a1T. Upon CL 0 CL;baseline dt CT;baseline QL normalization to Cyp , we obtain FC of mRNA expression þfLCLint;met;L 24a1T,baseline K of Cyp24a1. L

Cyp24a1 SM since Cyp24a1 ¼ 1 dð T Þ ð CT Þ T FC;L Cyp24a1T;baseline CT;baseline Cyp24a1T ¼ kin;Cyp24a1;T -kout;Cyp24a1;T dt Cyp24a1T;baseline Cyp24a1T;baseline QICP;baseline and CIð0Þ¼ CI;baseline ¼ ; QI þ fICLint;met;I KI or When fICLint,sec,I # QI (assumption is justified since the clearance of dCyp24a1 SMT 1,25(OH) D is low), FC;T ¼ CT ; 2 3 kin;Cyp24a1;T -kout;Cyp24a1;TCyp24a1FC;T dt CT;baseline QLCP;baseline CL;baseline simplifies to CLð0Þ¼CL;baseline ¼ since QL þ fLCLint;met;L KL ¼ ð Þ Cyp24a1T;baseline 1 A7 ðA12Þ For the mPBPK-PD model, the FC of Cyp24a1 (Cyp24a1 ) was FC,T For the Rate of Change in the Peripheral Tissues. described with the use of (k ) and (k ). in,Cyp24a1,T out,Cyp24a1,T dCBr CBr For the Rate of Change in Brain. VBr ¼ Q CP- dt Br KBr dAperi Cperi HBr ¼ Q C - ; C ð0Þ¼C ; ¼ K C ðA13Þ ; ; ð Þ peri P peri peri baseline peri P fBrCBrCLint met Br Cyp24a1FC;Br , where HBr is the power coefficient dt Kperi for the brain. After the discontinuation of 1,25(OH)2D3, levels of 1,25(OH)2D3 where Aperi is the amount of 1,25(OH)2D3 in the peripheral dCBr ¼ will rebound to basal conditions. After return to steady state, VBr dt compartment. 0 and CPð0Þ¼CP;baseline;

QBrCP;baseline The mPBPK(SFM) Model CBrð0Þ¼CBr;baseline ¼ since Cyp24a1 ; ¼ 1 QBr FC Br þ fBrCLint;met;Br KBr Additional equations are defined for the enterocyte and serosa ðA8Þ regions for the SFM: PBPK-PD Modeling of 1,25(OH)2D3 in Mice 207 dC C dCyp27b1 g2 en ¼ þ en HI; FC;K ImaxCK Ven kaAlumen Qen CP- fICenCLint;met;I Cyp24a1FC;I ¼ kin;Cyp27b1;K 1- g g2 þ 2 dt KI dt IC50 CK k ; ; Cyp27b1 ; ; Cyp27b1 ; ; ¼ 1 ð Þ¼ ¼ QenCP;baseline ; ¼ out Cyp27b1 K FC K FC K baseline Cen 0 C ; when Cyp24a1FC;I 1; en baseline Qen ð Þ þfICLint;met;I A19 KI ¼ ; ¼ ð Þ and Qen fQQI Ven fQVI A14 where CT is the 1,25(OH)2D3 concentration in tissue (kidney, liver, dC C ser ¼ ser ; ð Þ¼ ¼ ileum, or brain), and g1 and g2 are the corresponding Hill coefficient for Vser Qser CP- Cser 0 Cser;baseline KICP and dt KI Cyp24a1 and Cyp27b1, respectively. Emax and Imax are the maximum inductive and inhibitory FC, respectively. EC50 and IC50, the tissue ¼ ; ¼ ð Þ concentration that results in 50% of E and I , respectively, were Qser 1-fQ QI Vser 1-fQ VI A15 max max fitted with the initial estimates obtained from plotting FC against tissue

Here, fQ denotes the fraction of intestinal plasma flow perfusing the concentrations (Table 2). enterocyte compartment, and is identical to the fraction of intestinal Additional equations for full PBPK(SFM) model are shown pre- blood flow perfusing the enterocyte region. viously (see eq. A14–A16). The liver compartment receives plasma returns from the enterocyte Authorship Contributions and serosa regions. Equation A12 (liver compartment) is modified: Participated in research design: Ramakrishnan, Yang, Cao, Mager, Pang. Conducted experiments: Quach, Chow. dC C C Downloaded from V L ¼ Q C þQ en þQ ser L dt HA P en K ser K Performed data analysis: Ramakrishnan, Yang, Quach, Cao, Mager, Pang. I I Suggested and used of mPBPK-PD model for fitting: Cao C L HL; Wrote or contributed to the writing of the manuscript: Yang, Ramakrishnan, -QL -fLCLCLint;met;L Cyp24a1FC;L KL Quach, Mager, Pang. QLCP;baseline CLð0Þ¼CL;baseline ¼ ðA16Þ QL References þ fLCLint;met;L KL Bell NH (1998) Renal and nonrenal 25-hydroxyvitamin D-1alpha-hydroxylases and their clinical dmd.aspetjournals.org significance. J Bone Miner Res 13:350–353. The initial condition or the baseline plasma concentration of Boxenbaum HG, Riegelman S, and Elashoff RM (1974) Statistical estimations in pharmacokinetics. J Pharmacokinet Biopharm 2:123–148. 1,25(OH)2D3 [CP(0) or CP,baseline] was assigned the measured Brown RP, Delp MD, Lindstedt SL, Rhomberg LR, and Beliles RP (1997) Physiological pa- value (217 pM). The baseline concentration of 1,25(OH)2D3 in rameter values for physiologically based pharmacokinetic models. Toxicol Ind Health 13: 407–484. tissues was defined with respect to CP,baseline, as described in eqs. Cao Y, Balthasar JP, and Jusko WJ (2013) Second-generation minimal physiologically-based A8–A16. pharmacokinetic model for monoclonal antibodies. J Pharmacokinet Pharmacodyn 40: 597–607. The initial condition for the amount of 1,25(OH)2D3 in gut lumen

Cao Y and Jusko WJ (2012) Applications of minimal physiologically-based pharmacokinetic at ASPET Journals on June 27, 2016 was the i.p. dose administered (eq. A11). In contrast, the i.v. dose was models. J Pharmacokinet Pharmacodyn 39:711–723. administered to plasma compartment directly (eq. A6). Chen HS and Gross JF (1979) Estimation of tissue-to-plasma partition coefficients used in physiological pharmacokinetic models. J Pharmacokinet Biopharm 7:117–125. Chow ECY, Durk MR, Cummins CL, and Pang KS (2011) 1a,25-Dihydroxyvitamin D3 up- regulates P-glycoprotein via the vitamin D receptor and not farnesoid X receptor in both fxr(-/-) The Full PBPK-PD Model (Fig. 2) and fxr(+/+) mice and increased renal and brain efflux of digoxin in mice in vivo. J Pharmacol Exp Ther 337:846–859. The mass balance equations of the full PBPK-PD model for Chow ECY, Magomedova L, Quach HP, Patel R, Durk MR, Fan J, Maeng HJ, Irondi K, Anakk S, and Moore DD, et al. (2014) Vitamin D receptor activation down-regulates the small hetero- describing plasma and tissue 1,25(OH)2D3 disposition are similar to dimer partner and increases CYP7A1 to lower cholesterol. Gastroenterology 146:1048–1059. those for the mPBPK-PD model. However, inhibition of Cyp27b1FC,K Chow ECY, Quach HP, Vieth R, and Pang KS (2013) Temporal changes in tissue 1a,25- was fully accounted for in the full PBPK-PD model for inhibition dihydroxyvitamin D3, vitamin D receptor target genes, and calcium and PTH levels after 1,25(OH)2D3 treatment in mice. Am J Physiol Endocrinol Metab 304:E977–E989. of the endogenous synthesis rate (Rsyn) in kidney. The mass Clemens TL, Adams JS, Horiuchi N, Gilchrest BA, Cho H, Tsuchiya Y, Matsuo N, Suda T, balance equation for kidney compartment (eq. A9) was modified and Holick MF (1983) Interaction of 1,25-dihydroxyvitamin-D3 with keratinocytes and fi- as follows: broblasts from skin of normal subjects and a subject with vitamin-D-dependent rickets, type II: a model for study of the mode of action of 1,25-dihydroxyvitamin D3. J Clin Endocrinol Metab For the Rate of Change in Kidney. 56:824–830. Cong D, Doherty M, and Pang KS (2000) A new physiologically based, segregated-flow model to explain route-dependent intestinal metabolism. Drug Metab Dispos 28:224–235. dCK ¼ þ CK ‐ ; VK RsynCyp27b1FC;K QK CP- fKCKCLint;met;KCyp24a1FC;K Davies B and Morris T (1993) Physiological parameters in laboratory animals and humans. dt KK Pharm Res 10:1093–1095. Dayneka NL, Garg V, and Jusko WJ (1993) Comparison of four basic models of indirect Q C ; þ R ð Þ¼ ¼ K P baseline syn ¼ – CK 0 CK;baseline since Cyp27b1FC;K and Cyp24a1FC;K 1 pharmacodynamic responses. J Pharmacokinet Biopharm 21:457 478. QK Doherty MM and Pang KS (2000) Route-dependent metabolism of morphine in the vascularly þ fKCLint;met;K KK perfused rat small intestine preparation. Pharm Res 17:291–298. ðA17Þ Durk MR, Han K, Chow EC, Ahrens R, Henderson JT, Fraser PE, and Pang KS (2014) 1a,25- Dihydroxyvitamin D3 reduces cerebral amyloid-b accumulation and improves cognition in where Rsyn is the zero-order, net synthesis rate of endogenous mouse models of Alzheimer’s disease. J Neurosci 34:7091–7101. 1,25(OH) D mediated by Cyp27b1. Fan J, Chen S, Chow EC, and Pang KS (2010) PBPK modeling of intestinal and liver enzymes 2 3 and transporters in drug absorption and sequential metabolism. Curr Drug Metab 11:743–761. Instead of correlating Cyp27b1FC;K and Cyp24a1FC,T to the corre- Henry HL (2001) The 25(OH)D(3)/1alpha,25(OH)(2)D(3)-24R-hydroxylase: a catabolic or biosynthetic enzyme? Steroids 66:391–398. sponding tissue concentration of 1,25(OH)2D3, the relative expressions Holick MF (2009) Vitamin D status: measurement, interpretation, and clinical application. Ann of Cyp24a1 and Cyp27b1 were expressed using indirect equations with Epidemiol 19:73–78. Hollander D, Muralidhara KS, and Zimmerman A (1978) Vitamin D-3 intestinal absorption in sigmoidal Emax/Imax curve models, as shown below: vivo: influence of fatty acids, bile salts, and perfusate pH on absorption. Gut 19:267–272. g1 Hsu CH, Patel S, Young EW, and Simpson RU (1987) Production and degradation of calcitriol in dCyp24a1FC;T EmaxCT – ¼ k ; ; 1 þ renal failure rats. Am J Physiol 253:F1015 F1019. in Cyp24a1 T g g1 dt EC50 1 þ C Jones G, Strugnell SA, and DeLuca HF (1998) Current understanding of the molecular actions of T vitamin D. Physiol Rev 78:1193–1231. 2 ; ¼ kout;Cyp24a1;TCyp24a1FC;T Cyp24a1FC;T;baseline 1 Lemay J, Demers C, Hendy GN, Delvin EE, and Gascon-Barré M (1995) Expression of the 1,25- dihydroxyvitamin D3-24-hydroxylase gene in rat intestine: response to calcium, vitamin D3 and ðA18Þ calcitriol administration in vivo. J Bone Miner Res 10:1148–1157. 208 Ramakrishnan et al.

Lemire J (2000) 1,25-Dihydroxyvitamin D3–a hormone with immunomodulatory properties. regulation of CYP27B1 on synthesis and CYP24A1 on degradation. Br J Pharmacol 172: Z Rheumatol 59 (Suppl 1):24–27. 3611–3626. Li L, Gardner I, Dostalek M, and Jamei M (2014) Simulation of monoclonal antibody pharma- Sharma A and Jusko WJ (1996) Characterization of four basic models of indirect pharmacody- cokinetics in humans using a minimal physiologically based model. AAPS J 16:1097–1109. namic responses. J Pharmacokinet Biopharm 24:611–635. Mager DE, Wyska E, and Jusko WJ (2003) Diversity of mechanism-based pharmacodynamic Sheiner LB and Beal SL (1981) Some suggestions for measuring predictive performance. models. Drug Metab Dispos 31:510–518. J Pharmacokinet Biopharm 9:503–512. Nestorov IA, Aarons LJ, Arundel PA, and Rowland M (1998) Lumping of whole-body physi- Shinki T, Jin CH, Nishimura A, Nagai Y, Ohyama Y, Noshiro M, Okuda K, and Suda T (1992) ologically based pharmacokinetic models. J Pharmacokinet Biopharm 26:21–46. Parathyroid hormone inhibits 25-hydroxyvitamin D3-24-hydroxylase mRNA expression Norman AW, Nemere I, Muralidharan KR, and Okamura WH (1992) 1 beta, 25 (OH)2-vitamin stimulated by 1 alpha,25-dihydroxyvitamin D3 in rat kidney but not in intestine. J Biol Chem D3 is an antagonist of 1 alpha,25 (OH)2-vitamin D3 stimulated transcaltachia (the rapid hor- 267:13757–13762. monal stimulation of intestinal calcium transport). Biochem Biophys Res Commun 189: St-Arnaud R (1999) Targeted inactivation of vitamin D hydroxylases in mice. Bone 25:127–129. 1450–1456. Topilski I, Flaishon L, Naveh Y, Harmelin A, Levo Y, and Shachar I (2004) The anti- Pang KS (2003) Modeling of intestinal drug absorption: roles of transporters and metabolic inflammatory effects of 1,25-dihydroxyvitamin D3 on Th2 cells in vivo are due in part to the enzymes (for the Gillette Review Series). Drug Metab Dispos 31:1507–1519. control of integrin-mediated T lymphocyte homing. Eur J Immunol 34:1068–1076. Pang KS and Chow EC (2012) Commentary: theoretical predictions of flow effects on intestinal Wu G (1995) Calculating predictive performance: a user’s note. Pharmacol Res 31:393–399. and systemic availability in physiologically based pharmacokinetic intestine models: the traditional model, segregated flow model, and QGut model. Drug Metab Dispos 40:1869–1877. Pilari S and Huisinga W (2010) Lumping of physiologically-based pharmacokinetic models and a Address correspondence to: Dr. K. Sandy Pang, Department of Pharmaceutical mechanistic derivation of classical compartmental models. J Pharmacokinet Pharmacodyn 37: Sciences, Leslie Dan Faculty of Pharmacy, University at Toronto, 144 College 365–405. Quach HP, Yang QJ, Chow EC, Mager DE, Hoi SY, and Pang KS (2015) PKPD modelling to Street, Toronto, Ontario Canada. E-mail: [email protected] predict altered disposition of 1a,25-dihydroxyvitamin D3 in mice due to dose-dependent Downloaded from dmd.aspetjournals.org at ASPET Journals on June 27, 2016 Appendix VII – List of Publications

Peer-Reviewed:

1. Chow EC, Wang JZ, Quach HP, Tang H, Evans DC, Li AP, Silva J, and Pang KS (2016) Functional Integrity of the Chimeric Mouse Liver: Enzyme Zonation, Physiological Spaces, and Hepatic Enzymes and Transporters. Drug Metabolism and Disposition 44:1524-1534.

2. Ramakrishnan V, Yang QJ, Quach HP, Cao Y, Chow EC, Mager DE, and Pang KS (2016) Physiologically-Based Pharmacokinetic-Pharmacodynamic Modeling of 1α,25- Dihydroxyvitamin D3 in Mice: Comparison of Minimal and Full PBPK-PD Models. Drug Metabolism and Disposition 44:189-208.

3. Quach HP, Yang QJ, Chow EC, Mager DE, Hoi SY, and Pang KS (2015) PKPD Modeling to Predict Altered Disposition of 1α,25-Dihydroxyvitamin D3 in Mice Due to Dose-Dependent Regulation of Cyp27b1 on Synthesis and Cyp24a1 on Degradation. British Journal of Pharmacology 172:3611-3626.

4. Kim YC, Kim IB, Noh CH, Quach HP, Yoon IS, Chow EC, Kim M, Jin HE, Cho KH, Chung SJ, Pang KS, and Maeng HJ (2014) Effects of 1α,25-Dihydroxyvitamin D3, the Natural Vitamin D Receptor Ligand, on the Pharmacokinetics of Cefdinir and Cefadroxil, Organic Anion Transporter Substrates, in Rat. Journal of Pharmaceutical Sciences 103:3793-3805.

5. Chow EC, Magomedova L, Quach HP, Patel R, Durk MR, Fan J, Maeng HJ, Irondi K, Anakk S, Moore DD, Cummins CL, and Pang KS (2014) Vitamin D Receptor Activation Down- regulates the Small Heterodimer Partner and Increases CYP7A1 to Lower Cholesterol. Gastroenterology 146:1048-1059.

6. Chow EC, Quach HP, Vieth R, and Pang KS (2013) Correlation Between Temporal Tissue 1α,25-Dihydroxyvitamin D3 Treatment in Mice In Vivo. American Journal of Physiology. Endocrinology and Metabolism 304:E977-E989.

Under Review/In Preparation:

7. Quach HP, Hoi SY, Chen J, Bruinsma A, Groothuis GMM, Li AP, Chow EC, and Pang KS (2016) Vitamin D Deficiency Triggers Hypercholesterolemia that is Reversed Upon Treatment with 1α,25-Dihydroxyvitamin D3 and Vitamin D3 in Mice (under review).

8. Chow EC, Quach HP, Zhang Y, Wang JZ, Li AP, Silva J, Evans DC, Lai Y, and Pang KS (2016) The Chimeric Humanized Liver Model: Bile Acid Imbalance and Disruption of Liver Transporters and Enzymes (under review).

9. Quach HP, Bukuroshi P, Dzekic T, Magomedova L, Cummins CL, and Pang KS (2016) Vitamin D Analogs as Cholesterol Lowering Agents: Discrepancy Between In Vitro and In Vivo Potency (in preparation).

264

10. Yang QJ, Quach HP, and Pang KS (2016) Pharmacodynamic Effects of the Vitamin D Receptor: P-glycoprotein in Kidney and Brain, Cholesterol Lowering in Liver, and Calcium Absorption in Intestine (in preparation).

Conference Abstracts:

1. Yang QJ, Quach HP, and Pang KS (2016) PBPK-PD to Examine 1α,25-Dihydroxyvitamin D3 Concentrations and Vitamin D Receptor Gene Targets. 2016 AAPS Annual Meeting and Exposition, Denver, Colorado.

2. Quach HP, Hoi SY, Bruinsma A, Durk MR, Chow EC, Groothuis GMM, and Pang KS (2015) Vitamin D Deficiency is a Risk Factor for Hypercholesterolemia: Role of the Vitamin D Receptor in CYP7A1 Activation and Cholesterol Lowering. 2015 AAPS Annual Meeting and Exposition, Orlando, Florida.

3. Quach HP, Dzekic T, Magomedova L, Chow EC, Cummins CL, and Pang KS (2015) Discrepancy Between In Vitro and In Vivo Potencies of Vitamin D Analogs on Cholesterol Lowering. 2015 CSPS Conference, Toronto, Ontario.

4. Chow EC, Wang JZ, Tang H, Quach HP, Ng R, Silva J, Evans DC, and Pang KS (2014) Is the Chimeric Humanized Liver Mouse Model Ready to Predict Human Drug Metabolism In Vivo? 19th North American ISSX Annual Meeting, San Francisco, California.

5. Quach HP, Bruinsma A, Chow EC, Groothuis GMM, and Pang KS (2014) Human Liver Slices to Examine the Role of Vitamin D Receptor (VDR) on Cytochrome P450 7α-Hydroxylase (CYP7A1). 2014 Workshop on Vitamin D, Chicago, Illinois.

6. Quach HP, Yang QJ, Chow EC, Hoi SY, Durk MR, and Pang KS (2013) Pharmacokinetics of 1α,25-Dihydroxyvitamin D3 in Mice: Dose- and Route-Dependency. 2013 AAPS Annual Meeting and Exposition, San Antonio, Texas.

7. Ramakrishnan V, Quach HP, Cao Y, Chow EC, Pang KS, and Mager DE (2013) A minimal physiologically-based pharmacokinetic model of 1α,25-dihydroxyvitamin D3 in mice. 2013 AAPS Annual Meeting and Exposition, San Antonio, Texas.

8. Bukuroshi P, Quach HP, Chow EC, Cummins CL, and Pang KS (2013) Characterizing Novel Analogs of 1α,25-dihydroxyvitamin D3 for Vitamin D Receptor (VDR) Response. 2013 AAPS Annual Meeting and Exposition, San Antonio, Texas.

9. Chow EC, Quach HP, Vieth R, and Pang KS (2012) Changes in Plasma and Tissue 1α,25- Dihydroxyvitamin D3 During Repetitive Dosing in Mice: Their Correlation to Vitamin D Receptor Target Gene Changes. 2012 Workshop on Vitamin D, Houston, Texas.

10. Quach HP, Durk MR, Chow EC, and Pang KS (2011) Vitamin D Receptor (VDR) Effects of Lithocholic Acid Acetate (LCAa), an Alternate VDR Ligand, on Liver, Kidney and Brain of Mice Mimic Those of 1α,25-Dihydroxyvitamin D3. 2011 AAPS Annual Meeting and Exposition, Washington, D.C.

265